REACTORS, KINETICS, AND CATALYSIS

JOURNAL REVIEW

Monoliths as Multiphase Reactors: A Review

Shaibal Roy, Tobias Bauer, and Muthanna Al-Dahhan
Chemical Reaction Engineering Laboratory, Washington University, St. Louis, MO 63130

Peter Lehner and Thomas Turek

Bayer Technology Services, Bayer AG, D-51368, Leverkusen, Germany

DOI 10.1002/aic.10268
Published online in Wiley InterScience (www.interscience.wiley.com).

Monolith reactors are being studied as a replacement for conventional multiphase

reactors such as trickle-bed reactors, slurry reactors, and slurry bubble column reactors
for gas–liquid–solid reactions. Reactors with monolith catalyst packing have been found
to be hydrodynamically superior to existing industrial reactors. This review covers
multiphase reactions carried out in monolith reactors by various researchers. It first
defines the monolith reactor and looks into the geometrical aspects of monolith. The
section dealing with hydrodynamics reviews pressure drop, phase holdup, flow distribu-
tion, and dispersion characteristics. This study also considers the tools used to charac-
terize the hydrodynamic parameters and their typical values. Although the available
literature is scarce, monoliths are considered to have superior mass transfer character-
istics. This review lists the mass transfer correlations for each category (gas–liquid,
gas–solid, liquid–solid). The last section discusses the reaction aspects of monolith
reactors. The ultimate goal is to implement such reactors for multiphase reactions. This
section also compares the performance of monolith reactors with conventional multiphase
reactors and lists the various reactor models reported to predict the overall performance
of monolith reactors. © 2004 American Institute of Chemical Engineers AIChE J, 50: 2918 –2938,
2004
Keywords: monolith, structured packing, hydrodynamics, Review

Introduction Different types of reactors have been used for three-phase

Among the various chemical reactions occurring in broad
range of industrial application areas, catalytic gas–liquid–solid
reactions are widespread. These reactions occur extensively in
chemical, petroleum, petrochemical, biochemical, material,
and environmental industrial processes for a wide variety of
products (such as hydrogenation, oxidation, and alkylation).
Two notable examples are the catalytic hydrogenation of pe-
troleum fractions to remove sulfur impurities and the catalytic
oxidation of liquid hydrocarbons with air or oxygen (Leven-
spiel, 1996).
Correspondence concerning this article should be addressed to M. Al-Dahhan at
muthanna@che.wustl.edu.
© 2004 American Institute of Chemical Engineers
gas–liquid–solid reaction applications. The major ones are the
stirred tank slurry reactor, the slurry bubble column reactor,
and the packed-bed reactor. The choice is governed by the
reaction chemistry, the ease of use and manufacture of reactor
types, and the enhancement of mass transfer for mass transfer
limited reactions. Each reactor type has its own advantages and
drawbacks. Slurry catalysts are very small (micrometer scale,
5–50 ␮m), whereas packed-bed particles are larger (millimeter
scale, 1–3 mm). In general it can be stated that larger particles
are kinetically less efficient because of intraparticle diffusion
limitation. The solid–liquid slurry and the slurry bubble col-
umn reactors offer very simple reactor geometry, high heat
removal, excellent mass transfer characteristics, and a high
effectiveness factor because of the very small particle size. The
solid catalyst is suspended in the liquid medium through which

2918 November 2004 Vol. 50, No. 11 AIChE Journal

gas is dispersed. The major disadvantages of these reactors are
the separation of product and catalyst, and catalyst attrition.
The filtration of fine particles, needed to separate the catalyst
from liquid product, often makes this an inconvenient reactor
type. A packed-bed reactor, such as the trickle-bed reactor
(TBR), is much more convenient, although larger particles
must be used to guarantee moderate pressure drop. In these
reactors, gas and liquid flow cocurrently downward over the
catalyst. Liquid flowing in a bed tends to form channels and
bypasses—maldistribution of reactants is unavoidable. On the
catalyst surface, where the liquid is either depleted or imper-
fectly covers the catalyst surface, dry areas are encountered:
these substantially reduce the liquid–solid contacting efficiency
of the trickle-bed reactor. Moreover, local hot spots may de-
velop and cause runaways. Adding to the problem are the low
gas–liquid velocities required to avoid excessive pressure drop.
This requirement results in high operational costs and low
productivity. Another major drawback of conventional reactors
for multiphase reactions is the difficulty of scale-up to indus-
trial size units (Kapteijn et al., 2001).
To overcome the above-mentioned difficulties, research has
led to the use of structured packing instead of a random packed
bed.
There are different types of structured packing: packing
made of regular ceramic or metal support is called “monolith,”
whereas other types include “sandwich packing” and “open
cross flow structures.” The characteristic features of these types
of packing are uniform flow distribution, low pressure drop,
and enhanced mass transfer. The main advantage, and the
explanation of its noticeable performance enhancement, lies in
the manner how gas, liquid, and catalyst contact. As a result of
the application of regular structure, the scale-up to industrial
relevant size is considerably easier.
Through the application of these new structures, traditional
unit operations and reactors can be replaced by new, highly
productive and energy-efficient reactors. The new reactors will
provide more process security, less by-product, and reduced
dimensions for the same productivity. Stankiewicz and Mou-
lijin (2000) described these novel developments, which have
led to dramatic improvements in equipment and methods, such
as process intensification.
This article will consider only the use of monolithic catalyst
carrier for gas–liquid–solid reaction systems.
Monoliths have been successfully used for the abatement of
NOx and CO emissions from automobile engines and in mass
transfer operations such as distillation and absorptions (Cybul-
ski et al., 1998; Ellenberger et al., 1999).
However, their potential for use in multiphase reactions has
not yet been fully realized. For slurry, trickle-bed, and mono-
lith reactors, a number of experimental and theoretical com-
parisons have been made between typical parameters of cata-
lytic gas–liquid reactions (see, for example, Mazzaroni et al.,
1987; Nijhuis et al., 2001; Stankiewicz et al., 2001). These
studies clearly show the superior behavior of the monolithic
reactor over traditional three-phase reactors with respect to
productivity and selectivity. Currently, Akzo Nobel produces
hydrogen peroxide on a large scale using a monolith catalyst
for the hydrogenation of anthraquinone to the corresponding
hydroquinones (Albers et al., 2001). However, for successful
application uniform gas–liquid distribution is inevitable be-
AIChE Journal November 2004
cause redistribution of gas–liquid flow over the reactor length
is not possible.
This work reviews recent advances made in the use of
monoliths as multiphase flow reactors. The review starts by
defining the monolith structure, and then explores various
hydrodynamic parameters, such as flow regime, liquid holdup,
pressure drop, and dispersion. This is followed by a section
dealing with mass transfer characteristics. These parameters
are needed to predict the overall performance, design, and
scale-up of monolith reactor systems. The overall performance
is most frequently based on the intrinsic rate of model reactions
on monolith catalyst; thus, various kinetic measurements per-
formed on such catalyst supports at an intrinsic rate level are
reported as well. Finally, the section on reactor performance
reviews the performance of monolith reactor systems, combin-
ing the hydrodynamic and kinetic parameters.
We attempt to summarize what is known about hydrody-
namics, transport phenomena, and kinetics in monolith reactors
for three-phase applications. It is hoped that this will stimulate
additional research activities, which is need to further advance
our understanding of this reactor type.
Monolith Structure
The pace of research on monoliths as multiphase reactions
has picked up in the last decade. Although most of the studies
are presently confined to the laboratory and pilot scale, work on
the hydrogenation of alkylanthraquinones in hydrogen perox-
ide production has reached full-scale operation (Albers et al.,
2001; Schöön, 1989). This section describes the characteristic
features of the monolith structure for the use as multiphase
reactor packing at the laboratory scale.
Monolith structures are essentially a single structure with
many thin, vertical, parallel channels, separated from each
other by walls. The channels are usually rectangular, but tri-
angles, hexagons, or more complex geometries also exist. To
increase the surface area, internal fins can also be provided.
Furthermore, the fins have a stabilizing effect on the gas–liquid
flow and allow operation in countercurrent mode without
flooding (Lebens et al., 1997, 1999a,b,c). To enhance turbu-
lence inside monolith channels, patented monolith structures
have also been developed in which either corrugation has been
provided or channels have been interconnected for radial trans-
port (Cybulski et al., 1998). Obviously, hydrodynamics of such
specialty monoliths will be different from that of the straight-
channel monoliths. Such details are currently unavailable in the
open literature.
The number of channels per cross section, the “cell density,”
typically ranges between 100 and 1200 channels per square
inch (cpsi). The void fraction varies between 0.5 and 0.9 and is
frequently expressed as the open frontal area (OFA). Typical
values for the wall thickness range between 0.006 and 0.05 cm.
A monolith structure is characterized by the wall thickness and
cell density, which are independent of each other. Various
geometric parameters related to a monolith having a square
cross section are listed below and correspond to Figure 1.
Cell Density
1
n⫽ (1)
L2
Vol. 50, No. 11 2919

Figure 1. Cross-section of a single cell (not to scale).
Open Frontal Area
共L ⫺ t w兲 2
OFA ⫽ n共L ⫺ t w兲 2 ⫽ (2)
L2
Geometric Surface Area
共L ⫺ t w兲
GSA ⫽ 4n共L ⫺ t w兲 ⫽ 4 (3)
L2
Hydraulic Diameter
冉 冊
OFA 共L ⫺ t w兲
dh ⫽ 4 ⫽ (4)
GFA 4 to the channel diameter, so just a thin liquid film separates the
The parameters in Eqs. 1–3 are defined as: t, wall thickness; L,
length from one channel wall center to the other; R, fillet
radius.
Monoliths are industrially produced by extrusion of a paste
containing catalyst particles or by extrusion of a support on
which the catalyst can be coated (washcoating). A general
overview about characteristics, fabrication, and typical appli-
cations can be found in Williams (2001), Garcia-Bordeje et al.
(2002), and Gulati (1998). A previous work by Lachman and
Williams (1992) also gives a good introduction to monolith
production, raw materials, and its application in catalytic two-
phase processes.
Inside the monolith, the channels are separated by the chan-
nel walls, and therefore no radial mixing occurs. This case is
similar to an ideal plug-flow reactor, and therefore the reactor
yield will be high.
The operating mode and flow distribution have strong effects
on the performance of these reactors. The reactor can be
operated in batch and in continuous modes. In the batch mode,
liquid is continuously circulated through the monolith unit until
the desired conversion is met. In the continuous mode, the
liquid flows only once through the monolith core. There are
three different flow arrangements possible: cocurrent down-
flow, cocurrent upflow, and countercurrent flow. Different
types of distributors can be used to achieve the desired uniform
flow distribution over the monolith cross section. Literature
cites the use of spray nozzles, shower heads, and ejectors for
downflow (Broekhuis et al., 2001; Hatziantoniou et al., 1984;
2920 November 2004
Satterfield et al., 1977), and hole plates, glass frits, and static
mixers for upflow mode (Thulasidas et al., 1995b); still other
variations exist.
Several configurations of the monolith reactor setup are
found in the literature. In the “monolith froth reactor,” reactant
mixture in the form of a froth is passed upward in the reactor
(Crynes et al., 1995; Thulasidas et al., 1995b). In the “in-line
monolith reactor,” the monolith pieces are incorporated in the
pipes carrying reactants (Stankiewicz et al., 2001). In the
“mononolith loop reactor” (Heiszwolf et al., 2001a), and in the
“ejector-driven monolith loop reactor” (Broekhuis et al., 2001),
the gaseous reactant is sucked into the reactor by the liquid
flow, making a closed loop of gas flow. The basic element in
these reactors is the monolithic support; however, they vary
with respect to cell density and void fraction. They also vary in
hydrodynamics, such as in the direction of flow and recircula-
tion of the liquid and the gas phases.
In general, in multiphase reactors it is preferable to have a
high mass transfer rate to the catalyst surface. This rate can be
achieved by a high surface-to-volume ratio and a short diffu-
sion length of gas through the liquid. The monolith structure
provides a very high ratio, especially for monoliths with high
cell density. For example, a monolith with a cell density of 600
cpsi and an open frontal area of 82% has a surface-to-volume
ratio of 3476 m2/m3 (data provided by Corning, Inc.). A short
diffusion length can be obtained by operating the monolith in
two-phase slug flow (Taylor flow) or in the annular flow
regime. Taylor flow is characterized by a train of liquid slugs
and gas bubbles moving consecutively upward or downward
through the channel. The gas bubble length is several times the
channel diameter, and the gas bubble diameter is almost equal
gas from the wall. Typically, the liquid film thickness is in the
range of 30 to 70 ␮m (Irandoust et al., 1989a; experiments in
a circular glass capillary). The annular flow regime is charac-
terized by the flow of liquid along the channel wall as a thin
film and gas flow in the core of the channel. This regime can be
achieved for low liquid and high gas flow rates.
In summary, the primary advantages of the monolith packing
are low pressure drop in the channels and high mass transfer
rates attributed to small diffusion paths. Therefore, carrying out
equilibrium-limited reactions, such as hydrogenation and par-
tial oxidation in monoliths, is a promising proposition. More-
over, many hydrogenation reactions of industrial relevance are
consecutive reactions, and for such reactions, monoliths have
been shown to perform better with respect to productivity and
selectivity (Nijhuis et al., 2001).
Hydrodynamics
One of the most important criteria for selection of a mul-
tiphase reactor is choosing appropriate hydrodynamics
(Krishna et al., 1994). Apart from high conversion, factors such
as low pressure drop, high operating velocities without flood-
ing and instabilities, the extent of axial mixing, and the proper
distribution of reactant for effective utilization of catalyst are
part of the wish list for reactor design. A basic understanding
of hydrodynamics of monolith reactors is essential to their
design, scale-up, scale-down, and performance.
Researchers in this field range from experiments to develop
models for predicting the results a priori. Experimental studies
Vol. 50, No. 11 AIChE Journal
 channels, it is virtually impossible to determine the flow char-
acteristics and flow regime changes inside the monoliths.
Therefore, new noninvasive measurement techniques, such as
magnetic resonance imaging (MRI), can aid in depiction of
flow regimes inside monoliths. Two recent reports on the use of
MRI to determine slug flow were published by Mantle et al.
(2002) and Gladden et al. (2003). The extension of the appli-
Figure 2. A monolith channel operated in slug-flow re- cation to other flow regimes is desirable. Table 1 summarizes
gime (from Coleman et al., 1999). the work performed in single capillary tubes and monoliths.
The choice of the particular regime for industrial applica-
tions depends on the nature of reaction and the process condi-
have primarily been conducted in single circular capillary
tions. Flow regimes are mainly influenced by the gas and liquid
tubes, for ease of visualization, but have also been conducted in
properties, superficial velocities, and the diameter of the cap-
laboratory-scale monoliths (1–5 cm diameter). Researchers
illary and monolith channel.
have developed both simple semiempirical correlations as well
The overall gas–liquid holdup, pressure drop, and distribu-
as detailed numerical simulations, taking into account all the
tion inside the reactor depend on the flow regime in which the
possible physics.
reactor is working.
The hydrodynamic parameters such as flow regime, two-
Most of the authors have described the flow regimes as
phase flow pressure drop, and liquid-holdup mainly depend on
consisting of bubbly flow at low gas and high liquid velocities.
superficial liquid and gas velocities, physical properties of the
They have found slug flow, churn flow, and annular flow at
fluids, as well as on the type of structured packing used, such
high gas and low liquid flows, as shown in Table 1.
as random packing, monolith, static mixer, or any other struc-
Although the above regimes were observed in cocurrent
ture. Furthermore, it also depends on the mode of operation:
upflow, any difference in downflow regimes caused by the
whether it is operated cocurrently or countercurrently, upflow
difference in flow direction is small because of the relatively
or downflow, and so forth.
negligible influence of gravitational force in capillary tubes
This section reviews the following: the flow regime normally
(Cybulski et al., 1998).
encountered in small-diameter capillary tubes as well as in
In the real world of monoliths, it remains to be seen how
monoliths for a range of gas and liquid velocities; studies on
capillary analysis can be extended to monolith beds, consider-
pressure drop in monoliths and delineate the relevant correla-
ing the fact that the monolith channels are mostly noncircular
tions; and the gas–liquid holdup (most often, though, the pres-
in cross section and the surface roughness properties could be
sure drop studies are related to holdup calculations and axial
significantly different from those of glass tubes. It is question-
dispersion). A major concern of researchers dealing with mul-
able whether the exiting theories obtained from experiments in
tiphase reactors is the gas–liquid distribution inside the reactor
capillary tubes can be extended to monoliths. Further investi-
and the means to achieve uniform distribution. There is a
gations into the effects of capillary forces, caused by sharp
general disagreement among researchers regarding the degree
corners of noncircular channels on the two-phase flow pattern,
of axial and radial dispersion in monoliths. Some of the studies
are inevitable. The authors believe that only noninvasive mea-
reviewed in this section have found the dispersion to be sig-
surement techniques, such as MRI in conjunction with the
nificant.
consideration of the gas/liquid distribution device, can provide
Flow regime
A number of flow regimes have been identified in capillary
channels. These regimes are dispersed bubbly flow, slug flow
(Taylor flow) (see Figure 2), churn flow, and annular flow.
Primarily, these studies have been carried out by researchers
studying heat exchangers or in the field of nuclear engineering
(Coleman et al., 1999; Mishima et al., 1996; Triplett et al.,
1999a,b; Zhao et al., 2001). The common aim of these re-
searchers is to establish maps to visualize the flow pattern
inside the capillaries and to define the transition areas and flow
pattern maps using gas and liquid superficial velocities as
coordinates.
In a more recent study, Mishima et al. (1996) performed
experiments in capillary tubes. Figure 3 shows an experimen-
tally developed flow regime map for a glass capillary of 2.05
mm diameter, in which the model correlations developed for
large diameter tubes have been superimposed. Similar flow
maps from 1.05-, 3.12-, and 4.08-mm tubes are also available
Figure 3. Flow regime map in a single capillary (d ⴝ
(Mishima et al., 1996).
2.05 mm).
In the world of monoliths, similar flow regimes exist but
The open symbols indicate clearly identified flow regimes and
they are difficult to visualize. Because of the opaque nature of the solid symbols fall in the gray areas (from Mishima et al.,
the monolith material and the small hydraulic diameter of the 1996).

AIChE Journal November 2004 Vol. 50, No. 11 2921

 Table 1. Summary of Flow Regime Studies
Reference Structure and Flow Flow Range Reactor Dimensions Flow Regimes Studied
Scatterfield et al., Monolith with three Air–water L R ⫽ 122 cm Annular flow, slug flow
1977 different
distributors
Air–cyclohexane D R ⫽ 2.54 cm
u L ⫽ 0.33 to 6.58 cm/s n ⫽ 200 or 360 cpsi
Cocurrent downward u G ⫽ 0–150 cm/s Block length ⫽ 7.6 or
15.2 cm
Irandoust et al., 1989a Capillary glass tube Liquid loading ⫽ 0.04 L R ⫽ 200 mm Slug flow
to 0.43 m3/m2 s
Cocurrent downward u tot ⫽ 0.27 to 0.53 m/s d c ⫽ 1.5 mm
Mishima et al., 1996 Capillary tube Air–water d c ⫽ 1.05–4.08 mm Bubbly flow, slug flow,
churn flow, annular
flow
u L ⬃ 0.01–0.91 m/s
Cocurrent upflow u G ⬃ 0.09–80 m/s
(depending on d c )
Reineck et al., 1996 Monolith u L ⫽ 0–0.2 m/s LR ⫽ 2 m Bubbly flow, plug flow,
aerated slug flow,
annular flow
u G ⫽ 0–8 m/s D R ⫽ 118 mm
Cocurrent downward n ⫽ 300 and 400 cpsi
Lebens et al., 1997 Monoliths with n-Decane–air L R ⫽ 0.5 m Annular flow
internal fins and
beveled ends
u L ⫽ 0.009–0.05 m/s d c ⫽ 4.5 mm
Countercurrent u G ⫽ 0–5 m/s
Mewes et al., 1999 Monolith, TBR u L ⫽ 0.01–0.14 m/s L R ⫽ 30 mm Slug flow unaerated,
(without air) aerated
Cocurrent downward u L ⫽ 0.26–0.39 m/s D R ⫽ 120 mm
u G ⫽ 0–5 m/s d c ⫽ 2 mm
Coleman et al., 1999 Glass capillary u L ⫽ 0.01–10 m/s d h ⫽ 1.3 to 5.5 Bubbly, dispersed and
elongated bubbly
flow, slug flow,
wavy, annular wavy
and annular flow
Cocurrent vertical u G ⫽ 0.1–100 m/s
flow
Triplett et al., 1999a,b Glass capillary u L ⫽ 0.02–8 m/s d c ⫽ 1.1 and 1.45 Bubbly flow, churn
mm flow, slug flow,
annular flow
Cocurrent vertical u G ⫽ 0.02–80 m/s
flow
Zhao et al., 2001 Triangular glass u L ⫽ 0.1–10 m/s d h ⫽ 0.866, 1.443, Dispersed bubbly flow,
capillary and 2.886 mm slug flow, churn
flow, annular flow
Cocurrent upflow u G ⫽ 0.1–100 m/s

proper flow maps and transition criteria for monoliths. Unfor-
tunately, such a technique is costly, limited to reactor scale, and
only a few research groups work on this topic. Furthermore,
pressure drop measurements could give important information
about the flow regime and flow regime transitions inside the
monolith bed.
Moreover, the turbulent regime should be more thoroughly
investigated because it could significantly enhance the produc-
tivity at a pressure drop comparable to that of packed-bed
reactors. The factors influencing the instabilities in flow and
transition from one regime to another need to be thoroughly
investigated.
Pressure drop
Operations of multiphase processes, especially packed-bed
reactors, are always associated with pressure losses because of
the inner design of the reactors. Pressure drop represents the
energy dissipated caused by fluid flow through the reactor bed.
It is important in determining the energy losses, the sizing of
the compression equipment, liquid holdup, gas–liquid interfa-
cial area, and mass transfer coefficient (Al-Dahhan et al.,
1994). High pressure drop through the system not only requires
high energy input to the system, it also prohibits the unit from
being operated at high gas and liquid velocities, and thus the
throughput is limited. The use of novel structured catalystic
packing like monoliths addresses these concerns and reduces
investment and operation costs.
Starting from Ergun (1954), the estimation of pressure drop
along packed beds has been well documented over the years.
Besides a number of empirical correlations (Pinna et al., 2001)
phenomenological models have also been developed (Al-Dah-
han et al., 1997).
As mentioned earlier, one of the major advantages of struc-
tured packing is low pressure drop along the bed, which en-
ables the unit to run at higher capacity without encountering
hydrodynamic instability. Table 2 lists the pressure drop cor-

2922 November 2004 Vol. 50, No. 11 AIChE Journal

 Table 2. Pressure Drop Equations and Correlations for Capillaries and Monoliths
Author Pressure Drop Equation and Correlation
Standard pressure drop expression Frictional pressure drop per unit length of single phase:

冉 冊
⌬P
L f,i
⫽ 2f
␳i ui2
dh
(5)

Friction factor (Darcy, Fanning):

fD fF
f⫽ ⫽ (6)
8 2
For laminar flow, Re ⬍ 2100 (Hagen–Poiseulle law):
64 16
fD ⫽ fF ⫽ (7)
Re Re

冉 冊
⌬P
⌬L i
⫽
32␮␳i
d2h
共i ⫽ G, L兲 (8)

Satterfield et al. (1977) ⌬P m ⫽ ⌬P f ⫹ ⌬P or ⫺ g ␳ L L␧ L (9)

2
N共Vorif ⫺ Vtot
2
兲
⌬porif ⫽ 关␧L ␳L ⫹ 共1 ⫺ ␧L 兲␳G 兴 (10)
2
Grolman et al. (1996) ⌬P ⫽ ⌬P TP,f ⫹ ⌬P G (11)
64 L 1
⌬PTP,f ⫽ ␧L ␳ 共u ⫹ uG 兲2 (12)
ReTP dc 2 L L
⌬P G ⫽ Cu G with C ⫽ 45,000 Pa s⫺1 m (13)

Mewes et al. (1999) ⌬P uL uG ␳L ␧gG

⫽ ⫺␧L ␳L g ⫹ 32␮L 2 ⫹ 32␮G 2 ⫹ 共uL ⫹ uG 兲2 (14)
⌬L dC dC 2 Lb

␧gG 共1⫺␧ G 兲0.15

⫽ ␧ G ⱖ30% (15)
L b L b0 共1⫺0.15兲
␧gG ␧G
⫽ ␧G ⱕ 30% (16)
Lb Lb0

Heiszwolf et al. (2001a) Entry-region friction model:

冋
fTP ReL ⫽ 共f Re兲L ␧L 1 ⫹ 0.065 冉 冊 册 Lb
dc ReL
⫺0.66
(17)

Heiszwolf et al. (2001a)

冉 冊
⌬P
⌬L
1
⫽ ⫺fTP ␳L uTP
2
2
4
d
⫹ ␧L ␳L g (18)

F ␳L uTP dh
fTP ⫽ ReTP ⫽
ReTP ␮L

F ⫽ 18 (200 cpsi, OFA ⫽ 0.74)

F ⫽ 22 (400 cpsi, OFA ⫽ 0.75)
f ⫽ 28 (600 cpsi, OFA ⫽ 0.79)
Pseudo-homogeneous model:
f TP ReL ⫽ const

Lebens et al. (1999) dp

ⵜ共␮i ⵜUiZ 兲 ⫽ ⫹ g␳i (19)
dz
with i ⫽ G, L

relations reported by researchers working in the field of mono-
lith.
The pressure drop through the monolith is primarily caused
by factors such as (1) the wall friction, (2) the acceleration of
gas phase, (3) the orifice effect at the entry region and between
the monolith stacks, and (4) pressure drop caused by the
gas–liquid distributor. Mewes et al. (1999) also considered the
pressure drop attributed to aeration of liquid slugs, which
depends on the number of bubbles formed in the liquid slugs
(Table 2). If we assume the phases moving along the monolith
channel to be incompressible, the pressure drop caused by the
acceleration can be neglected.
In all the correlations listed in Table 2, the pressure drop
attributed to wall friction has been modeled in line with Hage-

AIChE Journal November 2004 Vol. 50, No. 11 2923

n–Poiseulle law. For specific monoliths, the friction factors
have been calculated experimentally as well as theoretically
(assuming laminar flow) with good agreement. In general, the
pressure drop attributed to the gas phase was found to be very
small.
Considerably less attention has been given to the last two
factors (orifice effects and distributor pressure drop). Satter-
field and Özel (1977) used the Bernoulli relationship to model
pressure drop resulting from orifice effects. However, the au-
thors found this contribution to be small compared to that of
friction losses.
Heiszwolf et al. (2001a) explained the phenomenon of bub-
ble formation in the liquid slugs and its effect on the pressure
drop. When gas bubbles in the channel are entirely contained
within the liquid and do not interact with the wall, the pressure
drop is high because of higher friction of liquid. When enough
gas is introduced, two counteracting effects take place. The
friction factor decreases because of gas friction with the wall
(which is less than liquid friction), but it also, to some extent,
reduces the laminar nature of the liquid slugs and thus increases
the friction factor. The authors proposed a two-phase friction
factor fTP to model the pressure drop.
Clearly, there is a greater need to investigate the entrance
and exit effects, particularly the role of gas bubbles of varying
size on the pressure drop. Researchers have used various types
of distributors in conducting experiments related to monoliths.
Most often, however, only the pressure drop across the mono-
lith has been reported, which could lead to undersizing of
pumps and compressors while designing a monolith unit.
Finally, it is important to investigate pressure drop in flow
regimes other than slug flow. In some of the recent investiga-
tion related to monolith, researchers are investigating the ap-
plication of monoliths in annular flow (Roy et al., 2002).
Holdup
Gas–liquid holdup is an important hydrodynamic parameter
for reactor design, scale-up, and performance modeling. Con-
siderable research has been performed on packed-bed and
trickle-bed reactors to determine phase holdup, and reviews are
available in the literature (see, for example, Al-Dahhan et al.,
1999). Several methods have been used to measure the holdup,
including gravimetric method (Nemec et al., 2001) for overall
holdup and the tomographic method for cross-sectional holdup
(Boyer et al., 2002; Chen et al., 2001). Determination of holdup
and holdup distribution is as important in structured beds as it
is in packed beds. The flow regimes significantly affect the
holdup. In structures such as sandwich beds and trickle beds,
the regimes are considerably different from those in monolith
because of the difference in flow paths. In monoliths, the
holdup is characterized by the formation of liquid slugs in the
channels in a regular fashion, whereas the liquid holdup in
trickle beds and other structured bed reactors arises from riv-
ulets between the particles and structures.
In monoliths operating in the Taylor flow (slug flow) regime,
the liquid holdup is simply the ratio of average liquid slug
length to the total length of the liquid and gas slug combined,
neglecting the contribution of liquid film surrounding the gas
bubble. This is true when the slugs do not coalesce, which is
the case reported by most investigators (Cybulski et al., 1998;
Grolman et al., 1996). Mewes et al. (1999) quoted previous
2924 November 2004
researchers in this field to predict that the ratio between the
average velocity of the liquid slugs and the average velocity of
bubbles differs only slightly from one. Thus, the average vol-
ume fraction of gas equals approximately the ratio of the flow
rate of gas to total flow rate, with a difference of about 15%. It
is therefore obvious that the proper measurement of the length
of the liquid and gas slug would give a fairly reliable measure
of the holdup.
Ishii (1977) investigated two-phase flow in round tubes and
developed the drift-flux (D-F) model to describe the relative
motion between phases in flow regimes. According to the D-F
model, the relationship between the gas velocity and the mix-
ture’s volumetric flux is expressed in Eq. 22 (Table 3). Later,
Mishima et al. (1996) provided a new distribution parameter
for two-phase flow in small diameter vertical tubes. This pa-
rameter was obtained from experiments in tubes with inner
diameters ranging from 1 to 4 mm, operated in bubbly flow and
slug-flow regimes.
Various techniques have been used to measure the lengths of
gas and liquid slugs. Hatziantoniou et al. (1984) and Irandoust
et al. (1988a) used two different techniques in a cocurrent
downflow mode. First, they measured the displacement of the
pump piston, which gives the liquid slug length, and measured
the frequency of the stroke for the gas bubble length. Second,
they used of a conductivity cell, whose response is shown in
Figure 4. Two probes were installed at the exit of a monolith
channel. One was placed at the center of the channel, the other
one at the channel wall.
The results obtained by the two methods were within 9.5%
of each other. It was also observed that the lengths of the gas
and liquid slugs could be independently varied by changing the
corresponding flow rates.
In subsequent works, Irandoust et al. (1989, 1992) used a
photocell to visualize the flow regime and to measure the plug
lengths in a glass capillary of diameter 2.2 mm. The lengths of
the gas slugs varied from 3.4 to 29.1 mm, and those of the
liquid varied from 2.9 to 67 mm, for a total average linear
velocity of 0.092 to 0.56 m/s.
Grolman et al. (1996) performed direct integral holdup mea-
surements. The liquid holdup was obtained by continuously
weighing the column. Furthermore, a model was presented that
takes into account the liquid slug and the liquid film between
the gas bubble and the wall. The experimental values are in
good agreement with the model prediction for liquid holdup
values higher than 65%. The authors assume that the discrep-
ancies at lower liquid holdup are caused by liquid maldistri-
bution.
Few researchers have considered holdup measurements in
flow regimes other than Taylor flow. The focus of their inves-
tigations has been the annular or film flow regime in counter-
current and cocurrent downflow mode. The annular flow re-
gime is akin to film flow: the liquid trickles down the corner of
the channels and the gas occupies the core.
Heibel et al. (2001a) successfully used magnetic resonance
imaging (MRI) to measure the holdup inside the monolith in
the film-flow regime. They used a cell density of 25 cpsi and
operated the monolithic reactor in downflow mode. Water
velocity was varied from 0.45 to 4 cm/s, and no gas flow was
used. Within the range of flow rates the liquid holdup varied
between 4 and 15%. The system was also modeled by solving
Navier–Stokes equations, assuming no slip between the gas
Vol. 50, No. 11 AIChE Journal
 Table 3. Holdup Correlations for Capillaries and Monolith Channels
Author (Year) Holdup Correlation

Wallis (1969) ␾G uG
␧G ⫽ ⫽ ⫽␤ (20)
共␾G ⫹ ␾L 兲 共uG ⫹ uL 兲

Butterworth (1975) 1 ⫺ ␧G
␧G
⫽A 冉 冊冉 冊冉 冊
1 ⫺ x p ␳G q ␮L r
x ␳L ␮G
(21)

For Lockart–Martinelli (1949) correlation:

A ⫽ 1.0, p ⫽ 0.64, q ⫽ 0.36, and r ⫽ 0.07
For Broncy (1963) correlation:
A ⫽ 1.0, p ⫽ 0.74, q ⫽ 0.65, and r ⫽ 0.13
Ishii (1977) v G ⫽ u G /␧ G ⫽ C 0 (u G ⫹ u L ) ⫹ V G,u (22)
uG
␧G ⫽ (23)
C0 共uG ⫹ uL 兲 ⫹ VG,u

Slug flow:
C 0 ⫽ 1.2 ⫺ 0.2 公␳ G / ␳ L (24)
⫺0.691d c
C 0 ⫽ 1.2 ⫹ 0.51e (Mishima et al., 1996) (25)
V G,u ⫽ 0.35 公⌬ ␳ gd c / ␳ L (26)

Grolman et al. (1996) uL ⫹ uG 共1 ⫺ R /R 兲 ⫺ ␾LF /␲R

2
0
2 2

␧L ⫽ (27)
共uL ⫹ uG 兲 ⫺ ␾LF /␲R2

and liquid interface, and the model’s predicted holdup was in
very good agreement with the MRI results except at low liquid
flow rates (Ul ⬍ 0.75 cm/s).
In general, the gas and liquid velocities inside the monolith
channels can be independently varied by changing the gas and
liquid flow rates into the system. Therefore the holdup can be
changed either by changing liquid or gas flow rates. Broekhuis
et al. (2001) and Heiszwolf et al. (2001a) described a monolith
loop reactor in which gas is recirculated within the system
using an ejector driven by liquid flow. The gas recirculation
velocity depends on the liquid velocity, the two being related
because of the equilibrium of pressure drop in the monolith
section and recirculation section. Therefore the gas holdup
cannot be independently changed and depends entirely on the
liquid velocity. Because the pressure drop in the two sections
Figure 4. Conductivity cell response at three different
gas/liquid flow rates.
(a) uL ⫽ 16.7 cm3/s, uG ⫽ 15.6 cm3/s; (b) uL ⫽ 16.7 cm3/s,
uG ⫽ 27.0 cm3/s; (c) uL ⫽ 10.8 cm3/s, uG ⫽ 15.6 cm3/s,
channel cross section 2.0 mm2, bed porosity 59% (from Hat-
ziantoniou et al., 1984).
also depends on the total pressure of the system, by changing
this pressure, the gas velocity and thus the gas holdup can be
changed to some extent.
Unfortunately, most of the research performed in this field
does not address the concern of holdup distribution across the
monolith cross section, which can prove to be an important
parameter for reactor design. Recently, noninvasive techniques
such as capacitance, X-ray and ␥-ray tomography, and MRI
have been successfully used in conventional reactors as well as
in structured beds (Harter et al., 2001; Heibel et al., 2001b;
Kumar et al., 1997; Mewes et al., 1999; Reinecke et al., 1998)
and can be extended to study the holdup distribution in mono-
lith packing.
Flow distribution
Uniform flow distribution in a multiphase reactor is impor-
tant for enhanced productivity and selectivity. It ensures com-
plete use of the catalyst and prevents hot spots in an exothermic
reaction system. For conventional reactors such as packed
beds, slurry columns, and bubble columns, distributions of the
gas and liquid phases have been studied in detail (Marcandelli
et al., 2000), including various techniques used to measure flow
maldistribution. The same measurement techniques have been
extended to a structured bed as well. The importance of gas/
liquid distribution inside monoliths is much more profound
because, unlike random packing or other structured packing,
once the liquid enters the reactor, there is no further redistri-
bution inside the monolith. Therefore the liquid must be dis-
tributed uniformly before it enters the monolith bed.
Several ways have been devised by various researchers for
liquid distribution in monoliths operating in gas–liquid cocur-
rent downflow mode. A showerhead was the most widely used
distribution device (Hatziantoniou et al., 1984, 1986; Irandoust,
1989; Nijhuis, 2001; Satterfield, 1977) because of its easy

AIChE Journal November 2004 Vol. 50, No. 11 2925

 Table 4. Summary of Distribution Systems Used by Different Researchers in Monolith
Direction
Author/Year of Flow Distributor Velocities Remarks
Satterfield et al. (1977) Downflow Shower head layer of u L ⫽ 0.3 to 6.6 cm/s Layer of monolith discs gave
spheres monolith reproducible pressure drop
discs values, indicating uniform
distribution
u G ⫽ 0–150 cm/s
Irandoust et al. (1989) Downflow Sieve plates u L ⫹ u G ⫽ 0.27 to
0.53 m/s
Crynes et al. (1995) Upflow Glass frit u L ⫽ 0.4 to 3,5 cm3/ Distributor produces froth,
s which then travels up
u G ⫽ 15.8 to 50 Overall monolith activity
cm3/s achieved was same as
intrinsic rate, indicating
uniform distribution
Reinecke et al. (1996) Downflow u L ⫽ 0–0.2 m/s Maldistribution observed
using capacitance
tomography. With
increasing liquid velocity,
distribution improved
u G ⫽ 0–8 m/s
Mewes et al. (1999) Downflow u L ⫽ 0.26–0.39 m/s Same as above
u G ⫽ 0–5 m/s
Broekhuis et al. (2001) Downflow Liquid-driver ejector u L ⫽ up to 20.4 cm/s The ejector action produces
fine gas bubbles in liquid

construction and operation. Table 4 gives a summary of distri-
bution systems used by different researchers.
Irandoust et al. (1989) evaluated the flow distribution in a
monolith operated in downflow by using several perforated
sieve plates, with perforations ranging from 5.7 to 27% open
area and hole diameter from 0.5 to 2.75 mm. The experiment
was performed in a column having a diameter of 25 cm. Water
flowed through the distributor, whereas air made a side entry
just before the monolith. It was observed that the liquid flow
distribution was dependent on pressure drop across the perfo-
rated plate. The liquid was poorly distributed when the pressure
drop across the plate fell below 100 –200 mm of water and
resulted in a much thicker spray of liquid over the monolith
channels. According to Cybulski et al. (1998), the liquid spray
drops should be much smaller than the channel diameter and
should be sprayed uniformly over the monolith.
Satterfield et al. (1977) investigated three different types of
liquid distributors for downflow arrangement and evaluated the
results for reproducibility. First, a flat distributor head with an
arrangement of 37 capillaries, varying from about 3 to 7 mm,
was used. The distribution was not uniform over the cross
section, which resulted in irreproducible pressure drop data. In
addition, a slight rotation of the head caused significant differ-
ences in pressure drop. The second distributor was a layer of
4-mm-diameter spheres. Some of the spheres plugged the en-
trance of monolith channels, and flooding above the layer
occurred at high liquid flow rates. This distributor also gave
unusable data. The authors obtained best distribution (indicated
by reproducibility of pressure drop) by using 27 randomly
aligned stacks of monoliths, each 3.2 mm thick, below the
showerhead.
Mewes et al. (1999) studied flow distribution in monoliths
by capacitance tomography. A spatial resolution of about
5–10% of the diameter of the measurement plane is in general
possible using capacitance tomography (Reinecke et al., 1999).
In this study, only liquid flow distribution in a 120-mm mono-
lithic reactor is considered. At low liquid velocities, the pres-
sure drop across the monolith is negative (hydrostatic pressure
is more than the frictional pressure drop), and this condition
forced recirculation of trapped gas inside the monoliths. With
increasing liquid velocity, the quality of liquid distribution
increases. The effect of gas velocity on the liquid distribution
was not demonstrated.
Crynes et al. (1995) developed a novel “monolith froth
reactor” in which a gas–liquid froth was made to flow up under
pressure. The froth was prepared by passing gas through a glass
frit (145- to 175-␮m pore diameter) over which liquid was
flowed. The system was tested for aqueous oxidation of phenol
in a ␥-alumina washcoat monolith impregnated with CuO as
active metal. The observed reaction rate was close to the
intrinsic rate, which the authors interpreted as a sign of good
phase distribution and minimal transport resistances.
Broekhuis et al. (2001) used a liquid-motive ejector as a
gas–liquid distributor. They claim that this ejector is also a very
good gas–liquid contactor, presaturating liquid before it enters
the reactor. It produces a fine dispersion of gas bubbles in
liquid, which results in excellent gas–liquid distribution over
the cross section of the monolith. The ejector also acts as a gas
compressor, resulting in higher superficial gas and liquid ve-
locities compared to those produced by gravity-driven mono-
lith reactors.
It is interesting to note that, in most of the studies mentioned
above, the role of gas velocities on the liquid distribution has
not been studied. In a critical observation of the published
works, the researchers have mainly depended on indirect meth-
ods of qualitatively estimating the flow maldistribution, such as
consistency in pressure drop measurement (Satterfield et al.,
1977) or high reactor productivity (Crynes et al., 1995). This
has resulted in no single parameter being used to quantitatively
define a maldistribution and compare the performance of these
distributor studied.
Modern-day noninvasive tools such as computed tomogra-

2926 November 2004 Vol. 50, No. 11 AIChE Journal

Figure 5. Axial dispersion coefficient for bubble-train flow in 2 mm.
(a) Circular capillary as a function of capillary number (Ca ⫽ ␮v/␴): E, experimental; f, theory. (b) Square capillaries: f, experimental data
(from Thulasidas et al., 1999).
phy, electrical and capacitance tomography, and MRI can help
generate quantitative images of the gas and liquid phase on a
cross section and longitudinal sections. These images will lead
us to better understand the dynamics of each type of distributor.
Axial and radial dispersion
Neglecting proper characterization of the mixing phenomena
can result in considerable error in the reactor performance. In
packed-bed reactors, the idealized assumption of plug flow for
modeling purposes is no longer pursued and considerable work
has been done to develop various levels of models such as the
axial dispersion model (ADM), tanks in series model, and so
forth (Ramachandran et al., 1983; Sundreasan, 1986; Taylor,
1953). These models attempt to account for the flow deviation
from plug flow character and to improve reactor performance.
In modeling monolith reactor performance, plug flow of the
liquid slugs has most often been assumed (Cybulski et al.,
1993, 1999; Edvinsson et al., 1994; Hatziantoniou et al., 1984;
Irandoust et al., 1989b,c; Stankiewicz, 2001). This assumption
is bolstered by the fact that in slug flow, the liquid slugs are
intensely mixed (Irandoust et al., 1989b) and there is very little
concentration gradient in either the radial or axial direction if
no chemical reaction takes place. Moreover, the thickness of
the liquid film around the gas bubbles is very small. There is
thus no interaction between two successive liquid slugs, and
hence axial dispersion is minimal.
In a study of residence time distribution (RTD) in monolith
froth reactor in cocurrent upflow, Patrick et al. (1995) observed
that interaction takes place as a result of dispersion in the thin
liquid film surrounding the gas slugs. A series of tracer studies,
by use of 4,6-dichlororesorcinol (DCR), was conducted in a
monolith reactor assembly with a cell density of 400 cpsi. The
actual liquid residence time in the monolith was calculated
from overall residence time measurements, using deconvolu-
tion by Fourier transform. When fitted into the experimental
RTD data obtained at a liquid flow rate of 4.7 cm3/s and a gas
flow rate of 47 cm3/s, a tank-in-series model predicted the
monolith as 1.15 perfectly mixed vessel, which indicates a high
degree of backmixing inside the monolith.
Thulasidas et al. (1995b) also reported modeling and exper-
AIChE Journal November 2004
imental work to determine the liquid-phase RTD in capillary
tubes of circular and square cross section, operated in upflow
mode. A corresponding mathematical model was developed,
based on the mass balance of the tracer element. In their model,
complete mixing within the liquid slugs and transport of tracer
through the liquid film surrounding the gas bubble were as-
sumed. The model and the experimental data were in agree-
ment for a circular capillary. It was found that the measured
mean residence time for the circular capillary was 20.0 s,
whereas the time predicted by the model was 19.6 s. The result
confirms the small amount of backmixing in circular capillar-
ies. However, the model failed to predict the mean residence
time for square channels as well as it did for the monolith froth
reactor. The experimentally obtained mean residence time for a
square capillary was 27.9 s. The authors argue that the liquid
slugs are not well mixed because the liquid flows through the
corners of the channels. The liquid flowing through the corners
of the channels will almost completely bypass the liquid slugs.
It is claimed that the model can be extended to the monolith
reactor by applying a statistical analysis of the flow within the
channels of the monolith.
Thulasidas et al. (1999) performed tracer studies in capillar-
ies with circular and square cross sections with a hydraulic
diameter of 2 mm and also studied a capillary bundle. The
bundle consisted of 96 square capillaries with a hydraulic
diameter of 2 mm. The studies were carried out in upward slug
flow mode. A mass transfer model was used to predict con-
centration vs. time curves for liquid slugs leaving the capillar-
ies. The results show good agreement between the model
prediction and the experimental data. Axial dispersion coeffi-
cients computed from experimental values of the Peclet num-
bers (Pe) for bubble train flow in circular and square capillaries
are shown in Figures 5a and 5b.
The theoretical model was extended for a bundle of capil-
laries and used to estimate residence time distributions. Nor-
malized concentration vs. time curves for the capillary bundle
are shown in Figure 6a; the computed RTD for a capillary
section is shown in Figure 6b. The average residence time from
the distribution shown in the latter figure is 69.41 s. The model
predicts well the RTD in a single capillary but for an extension
Vol. 50, No. 11 2927
Figure 6. (a) Experimental normalized concentration distribution for liquid-only flow and bubble flow in a capillary
bundle: f, liquid flow only; E, bubble train flow; (b) normalized concentration vs. time distribution for
capillary bundle after deconvolution; f, experimental tracer concentration vs. time distribution (from
Thulasidas et al., 1999).
to capillary bundles or monoliths it needs further studies of the
flow inside these arrays.
Obviously, there is an apparent disagreement among re-
searchers about the presence of axial dispersion in monoliths.
Even though cocurrent downflow operation has been assumed
to be plug flow, direct experimental verification of this assump-
tion is not available. In almost all studies, the regime of interest
was slug flow (Taylor flow). However, investigating peripheral
regions such as bubbly flow and churn flow will be of interest
for many mass transfer limited reactions. For annular flow, the
flowing film is considered laminar and treated accordingly
during modeling (Roy et al., 2002).
Mass Transfer
As mentioned earlier, one of the most attractive features of
monolith as compared to random packing for a multiphase
reaction is its enhanced mass transfer characteristics. In the
Taylor flow regime, an intense mixing takes place within the
liquid slug, which significantly enhances the liquid–solid mass
transfer rate. Moreover, the film separating gas plugs and
channel walls is very thin, posing minimal gas–solid mass
transfer resistance. Most of the correlations for liquid–solid
mass transfer developed so far in random packing are based on
power-law correlations relating the Sherwood number with the
Reynolds number, Schmidt number, and other geometrical
properties of mass transfer devices (Highfill et al., 2001).
Because the mechanism of mass transfer is expected to be the
same in monoliths, the same dimensionless groups may play
important roles in the development of mass transfer correla-
tions in monoliths.
The overall mass transfer rate in monolith depends primarily
on four different mass transfer phenomena, as depicted in
Figure 7. Because the reaction takes place on the surface and
inside the pores of the washcoated wall, the gas and liquid
reactants have to transfer from the bulk to the catalyst surface.
The gaseous reactant diffuses from the bulk of the gas slug to
the liquid bulk, and the corresponding flux depends on the
2928 November 2004
gas–liquid mass transfer coefficient (kGL). The transport of
liquid reactant and the dissolved gas in the liquid from the bulk
of the liquid to the washcoat surface depends on the liquid–
solid mass transfer coefficient (kLS). Gas in the gas slug also
diffuses to the surface of the catalyst through the thin liquid
film, which is characterized by the gas–solid mass transfer
coefficient (kGS). Finally, the reactants on the surface of the
catalyst have to diffuse into the pores of the washcoat or the
catalyst wall by standard diffusion processes. This diffusion
has been dealt with extensively in standard reaction engineer-
ing text books (Fogler, 1992; Froment and Bishoff, 1990;
Levenspiel, 1998) and will not be discussed in detail here. All
the above information is needed to adequately model the over-
all performance of the reactor. In subsequent sections, we will
explore each of the transport properties in detail.
Liquid–solid mass transfer
Hatziantoniou et al. (1982) studied the liquid–solid (L-S)
mass transfer in capillary tubes of 2.35 and 3.094 mm diame-
ters. The channel wall was coated with benzoic acid, and water
and air were passed cocurrently downward. The outlet concen-
tration of benzoic acid dissolved in water was measured. The
L-S mass transfer coefficient was extracted from experimental
results and by performing a differential mass balance on ben-
zoic acid assuming a plug-flow behavior. Based on the exper-
imental results, the author proposed the following correlation
for the liquid–solid mass transfer coefficient
Figure 7. Different transport phenomena in Taylor flow
inside a monolith.
Gas/solid (G-S), gas/liquid (G-L), liquid/solid (L-S), pore
diffusion.
Vol. 50, No. 11 AIChE Journal
 Sh ⫽ 3.51 冉 冊
ReSc
␥
0.44
␤⫺0.09
where ␥ ⫽ Lc/dc and ␤ ⫽ Ls/dc.
It is interesting to note that the expression contains the tube
length, which makes it unsuitable for scale-up. The flow rate at
which the expression was developed was significantly high so
as to maintain complete recirculation within the liquid plugs,
and thus the correlation cannot be used at low Reynolds num-
bers.
Based on a dissolution study in cylindrical capillaries, Iran-
doust and Andersson (1988) developed a new correlation to
estimate the liquid–solid mass transfer coefficient
Sh ⫽ 1.5 ⫻ 10⫺7 共Re兲1.648 共Sc兲0.177 共␣兲⫺2.338
where ␣ is the dimensionless film thickness, ␦f/dc. The authors
cited data from an unpublished work to formulate the above
correlation.
Bercić et al. (1997) carried out an experiment similar to that
of Hatziantoniou et al. (1982), in a glass tube of 2.5 mm
diameter coated with benzoic acid. Air and water were passed
through the tube in a controlled manner to form Taylor flow
inside the tube. Nonlinear regression was used to fit the exper-
imental data to a model equation to correlate the liquid–solid
mass transfer coefficient with the unit cell length (UCL, the
sum of one gas and one liquid slug length) and the cell velocity v
0.069v 0.63
k LSa ⫽
关共1 ⫺ ␧ G兲UCL ⫺ 0.105UCL␧ G兴 0.44
Based on mass transfer experiments, Heiszwolf et al.
(2001b) proposed a semiempirical liquid–solid mass transfer
model
冋
Sh ⫽ 3.66 1 ⫹ 0.152
⌿s
ReSc 冉 冊 册 ⫺0.423
where ⌿s is the dimensionless liquid slug length LS/d. Unfor-
tunately, the basis for such a correlation remains unpublished.
However, the authors compared the above correlation with that
proposed by previous researchers (such as Eqs. 28, 30, and 31).
The comparison shows that all the correlations more or less
exhibit the same trend and are in the same range. However, the
values predicted by Heiszwolf are slightly higher than those
predicted by other researchers. The authors noted that the
values predicted by Eq. 29 (Irandoust and Andersson, 1988a)
were not in the range predicted by other correlations.
In a recent article Kreutzer et al. (2001) reported heat trans-
fer studies using CFD simulation in a 1-mm-diameter tube and
developed a correlation relating the Nusselt number (Nu) as a
function of the dimensionless liquid slug length (⌿s), Reynolds
number (Re), and Prandtl number (Pr) within the range 1 ⬍ ⌿s
⬍ 16, 7 ⬍ Pr ⬍ 700, and 10 ⬍ Re ⬍ 400. By analogy, the
authors proposed the following correlation for the liquid–solid
mass transfer coefficient
冋
Sh ⫽ 20 1 ⫹ 0.003 冉 冊 册
⌿s
ReSc
⫺0.7
AIChE Journal November 2004
(28)
As is evident, most of the correlations are dependent on the
liquid slug length. However, there is no method available to
estimate the slug length a priori. The individual researchers
have based their experimental results on their own observations
of slug lengths, and may differ from one study to another.
Most of the experiments were conducted on a smooth sur-
face and in clearly defined slugs of liquid. However, in an
actual reactor, the liquid slugs may be aerated, giving rise to a
“disturbance” on the liquid–solid interface. Moreover, mono-
liths likely to be used in industrial settings will seldom have
smooth surfaces. These factors may cause deviations from
model predictions (Nijhuis et al., 2001). Table 5 summarizes
reported correlations for liquid–solid mass transfer coefficients.
(29) Gas–liquid mass transfer
Gas–liquid mass transfer in nonstructured trickle-bed reac-
tors has been studied in great detail by past researchers. Cor-
relations that cover wide ranges of Reynolds number (Re) and
Schmidt (Sc) numbers are available in the literature (for a
review, see Ramachandran et al., 1983). Although the mecha-
nism of transport is not expected to be significantly different in
structured packing, the work in this field is somewhat limited
and incoherent (Heiszwolf et al., 2001b).
Similar to their liquid–solid mass transfer work, Bercić et al.
(1997) proposed a dimensional correlation for the gas–liquid
mass transfer coefficient. Gas–liquid mass transfer coefficients
were measured by physical absorption of methane in water,
using capillary tubes of three different diameters (1.5, 2.5, and
3.1 mm). Methane and water formed the gas and liquid slugs,
(30)
respectively. The measured data were fitted into the proposed
expression, which was developed by assuming plug flow of the
liquid slug and then using nonlinear regression as follows
0.111v 1.19
k GLa ⫽ (33)
关共1 ⫺ ␧ G兲UCL兴 0.57
(31)
Because the expression does not contain any physical pa-
rameters of the gas and liquid used, it lacks generality and can
be used only for methane–water systems. Heiszwolf et al.
(1999) attempted to make the expression compatible with other
systems by adding the following correction factor
共k GLa兲 A ⫽ 共k GLa兲 B 冉 冊
DA
DB
n
(34)
where n is the scaling factor: n ⫽ 1 for film theory, and n ⫽ 0.5
for penetration theory (Kreutzer et al., 2001). However, this
modification has its own drawbacks and could not achieve the
desired results (Heiszwolf et al., 1999).
In earlier work, Irandoust and Andersson (1988a) used the
following correlation (Eq. 35) to compute kGL as a needed
parameter for modeling the overall performance of a monolith
reactor. The authors cited unpublished work to support their
claim
(32)
Sh ⫽ 0.41 冑ReSc (35)
Vol. 50, No. 11 2929
 Table 5. Available Correlations for Liquid–Solid (L-S) and Gas–Liquid (G-L) Mass Transfer
Correlation
Author (Year) Liquid–Solid Mass Tranfer Gas–Liquid Mass Transfer

Hatziantoniou et al. (1982) Sh ⫽ 3.51 冉 冊

ReSc
␥
0.44
␤⫺0.09
with ␥ ⫽ L c /d c and ␤ ⫽ L s /d c
Irandoust et al. (1988a) Sh ⫽ 1.5 ⫻ 10⫺7(Re)1.648(Sc)0.177(␣)⫺2.338 Sh ⫽ 0.41公ReSc
with ␣ ⫽ ␦ f /d c
共Sh ⫺ 1兲
冋 冉 冊册
1/3
Irandoust et al. (1992) 1
1 ⱕ Re ⱕ 400: ⫽ 1⫹ Re0.41
Sc1/3 ReSc
100 ⱕ Re ⱕ 2000: Sh ⫽ 1 ⫹ 0.724 Re0.48Sc1/3
0.069v0.63 0.111v1.19
Bercić et al. (1997) kLS a ⫽ kGL a ⫽
关共1 ⫺ ␧G 兲UCL ⫺ 0.105UCL␧G 兴0.44 关共1 ⫺ ␧G 兲UCL兴0.57

Heiszwolf et al. (1999) 冋

Sh ⫽ 3.66 1 ⫹ 0.152 冉 冊 册
⌿s
ReSc
⫺0.423
DA n
共kGL a兲A ⫽ 共kGL a兲B
DB 冉 冊
n ⫽ 1.0 for film theory
n ⫽ 0.5 for penetration theory (Kreutzer et al., 2001)
0.093␨⫺0.87
Lebens et al. (1999a) Sh ⫽ 1.04 ⫹
1 ⫹ 0.047␨⫺0.61
Lc DL
with ␨ ⫽ 2
dfM uLM
Kreutzer et al. (2001)
冋
Sh ⫽ 20 1 ⫹ 0.003 冉 冊 册
⌿s
ReSc
⫺0.7

In a more recent paper, Irandoust et al. (1992) assumed the
hemispherical caps of the gas bubble to be rigid spheres and
cited previous literature on mass transfer correlations for rigid
spheres to propose the following gas–liquid mass transfer co-
efficient
determined by measuring desorption by nitrogen gas flow of
oxygen from oxygen saturated water. The authors also devel-
oped a correlation by solving the convection– diffusion equa-
tion for a smooth laminar falling liquid film flow on a vertical
plane, as follows

共Sh ⫺ 1兲
冋 冉 冊册
1/3
1 0.093␨⫺0.87
1 ⱕ Re ⱕ 400: ⫽ 1⫹ Re0.41 (36) Sh ⫽ 1.04 ⫹ (38)
Sc1/3 ReSc 1 ⫹ 0.047␨⫺0.61

100 ⱕ Re ⱕ 2000: Sh ⫽ 1 ⫹ 0.724 Re0.48 Sc1/3 (37) with

Irandoust et al. (1992) performed the experiments in a single L cD L

capillary tube with three different liquids: water, ethanol, and
ethylene glycol. Air and the liquid were passed cocurrently
upward, and the outlet concentration of oxygen in the liquid
was measured for model validation. However, the model over-
estimated the experimental results by about 30% and therefore
Eq. 37 should be multiplied by a factor of 0.686 before it is
used elsewhere.
During the last decade, processes with two-phase cocurrent
upward or downward flow have been well investigated in
monolith reactors. Different researchers have provided a vari-
ety of experimental and theoretical data. Recently a new type
of structure, the internal finned reactor, was presented by
Lebens et al. (1997). This structure enables countercurrent
operation in the monolith. The monolith reactor operates in the
annular flow regime, and the fins help to stabilize the liquid
film flow. The mechanism of mass transfer is therefore signif-
icantly different from monoliths operated in the slug-flow
regime.
Lebens et al. (1999a) reported both experimental and pre-
dicted results on data for gas–liquid mass transfer coefficients.
In the experiment, the gas–liquid mass transfer coefficient was
␨⫽ (39)
␦ f,m
2
u L,m
where Lc is the tube length, DL is the diffusivity, ␦f,m is the
maximum film thickness, and uL,m is the maximum liquid
velocity.
It must be noted that the Sherwood number is a function of
the dimensionless tube length (␨), and for higher values of ␨ it
approaches the asymptotic value of 1.04. The experimental
values in general had good agreement with the above correla-
tion (Eq. 38), except at the inlet region of the tube, where the
correlation underestimates the experimental values by a wide
margin.
The area of contact for gas–liquid mass transfer has always
been evaluated by assuming hemispherical plugs. However, in
reality, the shape is more complex and varies with capillary
number (Cybulski and Moulijn, 1998).
Despite all the work performed to develop mass transfer
correlations, they have yet to be used effectively in models of
monolith reactor performance at pilot plant or industrial scales.
Nijhuis et al. (2001) used the mass transfer correlation devel-

2930 November 2004 Vol. 50, No. 11 AIChE Journal

oped by Irandoust et al. (1992) to model a pilot scale reactor.
However, the reactor performance predicted by the model far
overestimated the actual performance in experiments. The au-
thors argue that in actual reactors the resistance layer is much
thicker than that in model assumptions. Moreover, the wash-
coated channel walls are irregular, which would contribute to
the increase in mass transfer resistance.
As mentioned in the previous section, aeration of the slug
may introduce error in operation near the churn flow regime. A
comprehensive study is needed to account for all of these
factors, so that the correlations can be used in overall modeling
with some degree of confidence. Table 5, found at the end of
this section, lists the available correlations for liquid–solid and
gas–liquid mass transfer coefficients.
Gas–solid mass transfer
Researchers have evaluated the gas–solid mass transfer in a
simplistic manner. The gaseous reactant diffuses to the solid
surface through the thin film surrounding the gas slug. The
thin-film model was therefore used to determine the gas–solid
mass transfer coefficient (Cybulski et al., 1993; Edvinsson et
al., 1994; Irandoust et al., 1988)
D film between the gas bubbles and the catalyst surface, leads to
k G-S ⫽ (40)
␦f
Therefore, to determine the gas–solid mass transfer coefficient,
one has to evaluate the film thickness (␦f) of the liquid film,
which depends primarily on the surface tension of the liquid
and the liquid diffusivity (D). However, the film thickness is
not uniform around the periphery of the monolith channel, and
a mechanism to account for this nonuniformity needs to be
designed. It must be noted that experiments to directly deter-
mine the gas–solid mass transfer coefficient are not available in
the literature.
Reactor Performance
So far we have dealt with the hydrodynamics and the trans-
ports inside monolith packing being operated in a range of
operating conditions. These studies help us to understand the
basic characteristics of monolith packing under cold flow con-
ditions—with no chemical reactions. Many unit operations,
such as absorption, stripping, and distillation, can be modeled
better with these basic understandings.
The primary objective of this work is to review the effec-
tiveness of monoliths for gas–liquid–solid reactions. Currently,
the conventional reactors, such as slurry, fluidized-bed, and
packed-bed reactors have been used in industry to carry out
three-phase reactions. For a decade now, researchers have
shown that structured packing could be a viable alternative to
that for mass transfer limited reactions. However, to effectively
compete with conventional reactors, a wide range of reactions
need to be studied in structured packing, and an adequate
model to predict performance also needs to be devised.
This section covers work on these topics by various re-
searchers. It includes investigations of different test chemical
reactions (see the following subsection). The reactions were
chosen for their industrial relevance or for being intrinsically
fast or slow. Although it is well known that monolith structures
AIChE Journal November 2004
favor fast reactions, studies of intrinsic slow reactions also help
gauge the effectiveness of the monolith.
Most of the researchers, while evaluating the performance of
monolith, have compared it with that of conventional reactors.
There has been a general disagreement about the basis on
which such comparison should be made, as discussed in the
section on performance comparison. Finally, the models pro-
posed by investigators are outlined in the section on reactor
modeling. These models are based on the mass balance of the
reacting species, taking into account different mass transport
phenomena occurring inside the packing.
Test chemical reactions used
Monolith reactors have primarily been used to catalyze au-
tomobile engine exhaust. More recently, multiphase reactions
such as hydrogenation, and partial and complete oxidation
reactions, which are traditionally performed in slurry, fluidized,
or packed-bed reactors, are being tried in structured packing,
such as monoliths, sandwich or open cross-flow structures, and
foam structures.
The advantages of structured packing lie in the field of fast
reactions, which are mass transfer limited. The intense mixing
in the segmented slug flow, together with the very thin liquid
enhanced mass transfer. In studying the overall performance of
monolith reactors, the researchers have chosen a particular
reaction for a variety of reasons. Some reactions represent
industrially relevant processes or exemplify a group of similar
reactions; other reactions have well-known kinetics and sub-
stantial amounts of experimental and theoretical data available.
Most of the research efforts has centered on hydrogenation
reactions, one of the most important reactions in the petroleum
and fine chemical industries. These reactions are typically a
three-phase reaction over a catalyst surface, in which hydrogen
reacts with liquid in the presence of a solvent, which improves
heat removal. Another type of reaction, catalytic wet oxidation,
can be used to purify industrial and human wastewater. Such
processes are characterized by the reaction of oxidizable liquid
or solid with a gaseous source of oxygen, such as air or pure
oxygen, to produce CO2 and other innocuous end products
(Mishra, 1995). Table 6 gives a summary of research per-
formed using test reactions in monolith beds. As the table
suggests, most reactions studied in monolith reactors are either
hydrogenation or oxidation reactions, which are normally in-
trinsically fast reactions. This is the type of reactions monolith
is most suited for, given that enhancing mass transfer is what it
does best. It cannot enhance the intrinsic rate. On certain
occasions, very slow reactions have also been studied, to com-
pare the intrinsic reaction rate and overall reaction rate in a
monolith reactor.
Performance comparison between monolith and
conventional reactors
The focus of most investigations of monolith and other
structured packing for three-phase reactors is to evaluate their
performance vs. traditional reactors such as slurry bubble col-
umns, trickle-bed reactors, and batch reactors. Comparison has
been made both experimentally and by using model predic-
tions. Table 7 provides a summary of the comparison studies.
Frequently, the basis of comparison is the extent of conversion
Vol. 50, No. 11 2931
 Table 6. Summary of Reported Investigations Performed Using Test Reactions in Monolith Beds
Author Reaction Catalyst/Support Conditions Velocities/Mode of Operation
Hatziantoniou et al. (1984) Hydrogenation of nitrobenzoic acid 2.5 wt % Pd 310, 353 K u tot ⫽ 3.03–5.1 cm/s
1, 4 bar downflow
Hatziantoniou (1986) Hydrogenation of nitrobenzene and 5.4 wt % Pd 246, 376 K u tot ⫽ 1.7–4.2 cm/s
m-nitrotoluene SiO2 5.9, 9.8 bar downflow
Mazzaroni (1987) Hydrogenation of ␣-methylstyrene 1 wt % Pd 303–323 K u L ⫽ 0.05–0.34 cm/s
0.2–1 bar u G up to 0.12 cm/s upflow,
downflow
Irandoust (1988a) Hydrogenation of 2-ethylhexenal 0.5 wt % Pd 413, 433 K u tot ⫽ 2.3–8.5 cm/s
SiO2 4.0, 9.8 bar downflow
Kawakami (1989) Oxidation of glucose Glucose oxidase 298 K u L ⫽ 0.004–0.25 cm/s
u G ⫽ 0.5–6.0 cm/s upflow,
downflow
Irandoust (1990) Hydrodesulfurization of thiophene 12 wt % Co 509, 523 K u tot ⫽ 1.72–2.13 cm/s
and hydrogenation of 4 wt % Mo 30, 40 bar downflow
cyclohexene ␥-Al2O3
Edvinsson (1993) Hydrodesulfurization of C–Mo 543, 558, 573 K u tot ⫽ 4.6 cm/s downflow
dibenzothiophene ␥-Al2O3 60, 70, 80 bar
Edvinsson (1995) Hydrogenation of acetylene 0.04 wt % Pd 303, 313 K u L ⫽ 0.66 cm/s
␣-Al2O3 13, 20 bar u G ⫽ 6.5 cm/s downflow
␥-Al2O3
Crynes (1995) Oxidation of aqueous phenol CuO 383–423 K u L ⫽ 0.4–3.5 cm/s
␥-Al2O3 4.8–11.7 bar u G ⫽ 15.8–50 cm/s upflow
Smits (1996) Hydrogenation of a mixture of Pd 316, 343 K u tot ⫽ 5.0–45.0 cm/s
styrene and 1-octane in toluene ␣-Al2O3 5–15 bar downflow
Klinghoffer (1998a,b) Wet oxidation of acetic acid Pt 385–573 K u L ⫽ 0.024–0.093 cm/s
Al2O3 3.5–128 hour u G ⫽ 2.36 cm/s (SCM)
upflow
Patrick (2000) Wet oxidation of glucose and 0.26 wt % Pt 333–455 K u L ⫽ 0.024 cm/s
cellulose Al2O3 u G ⫽ 1.48 cm/s (SCM)
upflow
Nujhuis (2001) Hydrogenation of ␣-methylstyrene 1 wt % Ni 373, 423 K u tot ⫽ 20 cm/s downflow
hydrogenation of benzaldehyde 9 wt % ␥-Al2O3 10, 15 bar
Schutt (2002) Wet oxidation of cellulose 0.34 wt % Pd 373–428 K u L ⫽ 0.026 cm/s
Al2O3 u G ⫽ 2.2 cm/s (SCM)
upflow
Liu (2002) Dehydrogenation of ethylbenzene 72 wt % 866, 878 K u L ⫽ 0.0014 cm/s downflow
␣-Fe2O3 1 bar
16 wt % K2O
4 wt % CeO2

under similar operating conditions, productivity, selectivity, ever, the mass of catalyst, mass of active metal, and external
and pressure drop. The volume of catalyst was the primary surface area have also been considered in the comparisons.
weighting factor for comparison of different reactors. How- Compared to the conventional reactors, structured packing

Table 7. Summary of Studies on the Comparisons between Monolith and Conventional Reactors
Author
Mazzaroni et al. (1987)
Cybulski et al. (1993)
Edvinsson et al. (1994)
Cybulski et al. (1999)
Nijhuis et al. (2001)
Stankiewicz et al. (2001)
Heiszwolf et al. (2001a)
Roy (2002)
Comparison Kind of Study Reaction
Monolithic reactor vs. trickle-bed Experimnental studies Hydrogenation of ␣-
reactor methylstyrene
Monolithic reactor vs. literature Mathematical model of Liquid-phase methanol
data of slurry columns, monolith reactor synthesis
autoclaves, and trickle-bed
reactors
Monolithic reactor vs. trickle bed Numerical simulations Three-phase hydrogenation
Monolithic reactor vs. agitated- Mathematical modeling of Hydrogenation of 3-
slurry reactor both reactors hydroxypropanal
Monolithic reactor vs. trickle-bed Pilot-scale study Hydrogenation of ␣-
reactor methylstyrene and
benzaldehyde
In-line monolith reactor vs. Theoretical studies
trickle-bed reactor
Monolith loop reactor vs. bubble Theoretical studies
column and slurry reactor
Generic method to design Theoretical studies
monolith catalyst
Monolith reactor vs. pellet-based
trickle-bed reactor

2932 November 2004 Vol. 50, No. 11 AIChE Journal

leads to differences in catalyst loading, mass transfer resis-
tance, contacting area, and pressure drop because of the regu-
larity. There is no doubt that the new structures have a superior
advantage with respect to pressure drop. The other parameters,
however, have to be investigated further and optimized to
guarantee a better overall performance than that of traditional
reactors. Unfortunately, various investigators have used differ-
ent criteria for the comparison of monoliths with traditional
reactors.
Mazzarino and Baldi (1987) investigated hydrogenation of
␣-methylstyrene into cumene using a ceramic monolith coated
with Pd. Comparison was made with a trickle-bed reactor
(TBR) based on catalyst weight, and the monolith reactor was
found to perform better than the TBR at high gas flow rates,
whereas the reaction rate was virtually insensitive to the liquid
flow rate. The monolith performed better in upflow mode than
in downflow mode because of better wetting of the catalyst.
Figure 8 shows the hydrogenation rates per unit Pd mass of
␣-methylstyrene in the monolith reactor and the TBR at dif-
ferent gas flow rates under the same operating conditions. It
should be noted that better mass transfer in monolith is attrib-
uted to recirculation within the liquid slugs and to the thin
liquid film separating the gas slug from the catalyst layer, as
mentioned earlier. To demonstrate this point, the authors con-
ducted a single-phase reaction, in which the liquid was satu-
rated with hydrogen before entering the reactor. In this circum-
stance the TBR performed better than monolith in terms of
overall conversion.
Cybulski et al. (1993) developed a mathematical model for
liquid-phase methanol synthesis. Different process conditions
and design parameters were simulated. It was found that the
optimum thickness of the catalyst layer is in the range of 50 –75
␮m. The authors found that the performance of a monolith
reactor is not always superior to that of a conventional reactor.
The monolith reactor’s performance was better with respect to
CO conversion and methanol productivity per unit volume of
the reactor when a stoichiometrically balanced syngas was used
as feed. The improvement stems from the fact that the monolith
reactor used contained a theoretically higher concentration of
catalyst than did the slurry reactor. In the case of the TBR, even
though the catalyst concentration is higher, the intraparticle
diffusional resistance is significantly higher.
Figure 8. Conversion rates per unit palladium mass of
hydrogenation of ␣-methyl styrene vs. gas ve-
locity at 313 K (from Mazzarino et al. (1987).
AIChE Journal November 2004
Figure 9. Attainable selectivity and production for
MASR, MR, and TBR (from Cybulsky et al.,
1999).
Edvinsson et al. (1994) used numerical simulation to quan-
titatively compare the performance of a conventional trickle-
bed reactor and a monolith reactor, based on the catalyst load.
The monolith reactor had washcoated catalyst with varying
washcoat thickness (resulting in catalyst fraction between 6 and
33%), whereas catalyst particles used in the simulated TBR had
eggshell with a thickness corresponding to approximately the
same catalyst loading (tshell values of 0.5 to 0.05dp). The model
reaction was a consecutive three-phase hydrogenation. For the
model system, the selectivity of the intermediate product was
found to be higher for the monolith over almost the whole
range of variables studied. For particles larger than 2 mm, the
volumetric space–time yield was higher in the monolith reactor
compared to that in the TBR. For particles smaller than 2 mm,
however, the pressure drop in the TBR is considerably higher.
The pressure drop in the MR was sufficiently low that gas
could be recirculated without the requirement of external
pumping.
Cybulski et al. (1999) conducted a comparative study of
hydrogenation of 3-hydroxypropanal to 1,3-propanediol in a
mechanically agitated slurry reactor (MASR) and a monolith
reactor. A purely numerical study was done to compare the
performance of the MR and MASR based on productivity
(mol/m3 s⫺1) and selectivity. Although the comparisons were
made on the basis of per unit reactor volumes, the authors
emphasized the point that much more catalyst can be packed in
an MR than in an MASR per unit volume. They adopted a
modeling approach similar to that used by Edvinsson et al.
(1994), discussed earlier. Intrinsic kinetics, mass transfer cor-
relations, and model equations derived earlier by various re-
searchers (Andersson, 1998; Zhu, 1995) were directly used in
this study. As shown in Figure 9, the numerical studies clearly
demonstrated the superior performance of the MR over MASR,
with respect to reactor productivity and process selectivity.
It should be noted that the actual productivity of a MASR is
higher than that shown in the figure. However, given that an
MASR requires considerable downtime for filling, emptying,
and filtering, the overall productivity decreases significantly.
Nijhuis et al. (2001) did a comparative study of hydrogena-
tion of ␣-methylstyrene and benzaldehyde in a trickle-bed
reactor and a monolith. Experimental results showed that the
Vol. 50, No. 11 2933
monolith reactor had a 50% higher activity per unit reactor
volume than that of the trickle-bed reactor (Table 8), even
though the porosity of the monolith reactor is relatively high—
this despite the fact that the intrinsic rate for the extruded
catalyst used in the TBR was slightly more than that of the MR
catalyst. This finding clearly demonstrates the fact that the
better performance in the MR is entirely attributed to its struc-
ture. Experiments also revealed that the enhanced MR perfor-
mance resulted from the fact that Taylor bubbles in an MR
form a thin film through which most of the mass transfer takes
place. If, instead, the liquid reactant is saturated with hydrogen
and the resultant single phase is passed through the MR, the
activity falls to about one fourth of the three-phase Taylor flow
regime.
MRs are also known to exhibit higher selectivity because of
the narrow distribution of their residence time. The hydroge-
nation of benzaldehyde showed significantly higher selectivity
(⬎90%) with respect to benzylalcohol in an MR, whereas a
TBR gave a selectivity of 73%, thus reaffirming the advantage
of the MR over the TBR.
Stankiewicz (2001) proposed a novel concept in which the
conventional reactor vessel is replaced with monolith units,
installed in the process pipelines. The reactor configuration is
known as an “in-line monolith reactor” (ILMR). The author
made comparison studies for a slow hydrogenation reaction
(k ⫽ 3.94 ⫻ 10⫺3 m3liquid s⫺1 kgcatalyst) in a monolith and
trickle-bed reactor with 3.5-mm impregnated catalyst particles.
The monolith was assumed to be operated purely in liquid
phase, the feed being saturated with hydrogen, much in excess
of stoichiometric proportions. To achieve 95% conversation of
the limiting reactant, the simulation results showed that for a
cell density ranging from 200 to 1100 cpsi and a washcoat
thickness of 17–70 ␮m, the required reactor volume is 6 to 53
times less than that of a conventional trickle-bed reactor. The
author also compared the performance of different cell densi-
ties and washcoat thicknesses and found that the required
monolith reactor volume increased with increasing washcoat
thickness and cell density.
Heiszwolf et al. (2001) used the conventional way of com-
paring the performance of different types of conventional re-
actors, while studying the hydrodynamic aspects of a monolith
loop reactor. The authors used the power input per unit volume
of the system (P/V) to obtain a certain gas–liquid mass transfer
rate as the indicator for reactor performance. The P/V for
monolith reactor was calculated using the following expression
P ⌬P L␧u L
⫽ (41)
V L
Table 8. Bed Characteristics Used in Comparing the
Performance of a Monolith and TBR Using Mass Transfer
Limited Hydrogenation of ␣-Methylstyrene over an Eggshell
Nickel Catalyst*
Monolith Trickle
Activity per Parameter Unit (400 cpsi) Bed
Reactor volume 3
mol/mreactor /s 6.2 4.6
Geometrical surface area 2
mol/mgeom /s 1.9 ⫻ 10⫺3 2.0 ⫻ 10⫺3
Catalyst amount mol/gcatalyst/s 1.1 ⫻ 10⫺5 6.2 ⫻ 10⫺5
Nickel amount mol/gnickel/s 9.8 ⫻ 10⫺4 8.4 ⫻ 10⫺5
*Conditions: 373 K, 10 bar H2 (from Nijhuis et al., 2001).
2934 November 2004
For stirred tanks and bubble columns, previously reported
correlations for P/V were used (Schluter et al., 1992; van’t Riet,
1979). For low P/V, It was found that the kGLa value for the
monolith configuration was much higher than that for other
conventional reactors. For high P/V (⬎10,000 W/m3), all types
of reactor give comparable kGLa values. It was further observed
that for monolith reactors, the kGLa value is a weak function of
P/V, meaning the mass transfer coefficient remains almost
constant for significantly high gas and liquid flow rates.
Reactor modeling
Because of its well-defined structure, the modeling of mono-
lith is considered more straightforward than the modeling of
random packed beds. However, the environment outside the
monolith (distributor, outlet arrangements) affects its perfor-
mance to a great extent. This section deals with different
models reported in the literature to simulate the performance of
monolith reactors. In most published work related to monolith,
maximum emphasis has been given to mass transfer correla-
tions and pressure drop because these packings have been used
in industry for over two decades as mass transfer elements,
such as in distillation and absorption.
Table 9 lists the models for monolith reactors reported in the
literature. Three distinct reactor models for monolith reactors
have been published in the open literature. Hatziantoniou et al.
(1984) and Irandoust et al. (1988a) proposed models for mono-
lith reactors operating in the Taylor flow regime. The one
proposed by Irandoust gained more attention and was subse-
quently used by other researchers (Edvinsson et al., 1994;
Stankiewicz et al., 2001). More recently, Liu et al. (2002) and
Roy et al. (2002) dealt with monolith reactors operated in the
film-flow regime, which is more akin to the trickle-flow regime
in packed beds. In the model developed by Irandoust et al.
(1988), and later adopted by Edvinsson et al. (1994) and others,
plug flow was assumed in the liquid phase. Three different
mass transfer fluxes were considered in the differential mass
balance for reacting species, as shown in Figure 10.
In all the studies involving this model, a good agreement was
observed between experimental studies (Irandoust et al., 1988)
and a superior performance compared to trickle-bed reactor
(TBR). However, Nijhuis et al. (2001) reported experimentally
determined activity of monolith per unit volume severalfold
lower compared to that predicted by using the model. The
authors cited reasons including irregular wall structure and
nonuniform liquid film thickness, both contributing to lower
mass transfer rate in the experiment. Several other factors that
could affect the performance of monolith are aeration of liquid
slugs (Mewes et al., 1999), irregular liquid and gas slug length
(Sederman et al., 2003), initial maldistribution of gas and liquid
reactants (Reinecke et al., 1996), and compression of gas slugs.
Important parameters for monolith reactor modeling, phase
holdup, and contact surface area for mass transfer depend on
these factors. A detailed study of these hydrodynamic effects is
the current need. Carefully executed experiments and a detailed
computation fluid dynamic (CFD) study could help us to better
understand the occurring phenomena.
Hatziantoniou et al. (1984) modeled hydrogenation of nitro-
benzoic acid (NBA) in a monolith reactor. The model was
simplified by the fact that the intrinsic reaction rate was zero
order with respect to NBA, and first order with respect to
Vol. 50, No. 11 AIChE Journal
 Table 9. Summary of Overall Reactor Models Available in the Literature
Author Flow Regimes Basic Approach Design Equation in the Bulk Interface Equations Hydrodynamic Equations
Edvinsson
et al.,
1994
Gas–liquid slug
flow
Three distinct
mass transfer
rates, plug
冋 册 冋
d FG
dz FL
⫺NGL ⫺ NGS
⫽ N ⫺N
GL LS
册 Interface between Plug flow in both gas
and liquid slugs, no
slip between gas and
flow in each liquid slugs
phase
Refer to Figure 10 Gas/liquid

NGL,i ⫽ 共kGL a兲i 冉Pi

Hei
⫺ cL,i 冊
Liquid/solid
NLS,i ⫽ 共kLS a兲i 共cL,i ⫺ cs,i 兲

⫽ fw xcat 冘
j
␯i,j ␩j rL,j

Gas/solid
NGS,i ⫽ 共kGS a兲i 共cI,i ⫺ cs,i 兲

⫽ 共1 ⫺ fw 兲xcat 冘
j
␯i,j ␩j rI,j

Roy et al., Annular flow (gas dc Within washcoat (to calculate Gas core/liquid film
2002 core/liquid uL0 ⫽ ⫺rapp catalyst effectiveness, ␩ e (annular) regime. Both
dz
film) liquid and gas under
laminar flow with no
slip at the interface.
r app ⫽ ␥ e ␩ i (1 ⫺ OFA)r cat,v ⵜ(D e ⵜc s ) ⫽ r cat,v ; c s ⫽
c bulk at S-L interface
ⵜc s ⫽ 0 at the plane of
symmetry within the solid
wall
Within liquid
⭸c
uL,z ⫽ ⵜ共Dⵜc兲
⭸z

hydrogen. This model differs from the model proposed by
Irandoust et al. (1988) only in that the primary mass balance of
hydrogen was carried out inside the washcoat as an unsteady-
state diffusion–reaction problem, instead of in the channel. The
transport phenomena in the channels acted as boundary condi-
tions for the design equation. Unfortunately, the model also
contains liquid and gas slug time constants, which cannot be
known a priori without doing actual experiments. For this
reason, and also because of the model’s inability to incorporate
more complex reaction rate forms, it has not gained favor in
subsequent years, and therefore is not included in Table 9.
Roy et al. (2002) published a part of their work on reactor
modeling of monolith reactors working in the film-flow regime.
The authors aimed to develop criteria to determine the opti-
mum design (washcoat thickness, channel diameter) of a
monolith based on the intrinsic kinetic rate of reactions. The
film-flow regime is achieved when the reactor is operated at
very low liquid velocity (0.1 to 2 cm/s), similar to trickle flow
in a packed-bed reactor. The authors proposed three levels of
modeling. A one-dimensional, plug-flow, steady-state, convec-
tion–reaction equation was advanced to describe the annular
film. It gives the concentration profile of the reacting species
along the length of the reactor. Two extra parameters, the
velocity profile in the film and the effectiveness factor arising
out of diffusion resistance inside the catalyst, are needed to
solve the above equation. They were modeled separately. To
obtain the effectiveness factor, the standard equation for slab
geometry ␩ ⫽ tan ␾/␾ was used. The Thiele modulus ␾ ⫽
l公(k/D) was found to be adequate for all shapes when the
diffusion length scale “l” was expressed as l ⫽ (1 ⫺ OFA)/
GSA. For each velocity, the velocity profile inside the film was
solved using the standard Navier–Stokes equation.
The results show that for slow reactions a monolith with high
solid fraction is preferred because the higher diffusion resis-
tance does not have a strong bearing and more catalyst can be
packed in a unit volume. For fast reactions, optima exist with
respect to the diffusion length, and any further increase in
catalyst wall thickness decreases the volumetric activity.
This latest study heralds a new question in the research
pertaining to monoliths. To date, slug flow has been advocated
as being the most advantageous flow regime. A comparison of
monolith performance operating in slug-flow and film-flow
regions is in order.

Figure 10. Different mass transfer fluxes used in the Acknowledgments

monolith reactor modeling of Irandoust et al. The authors gratefully acknowledge the financial support provided by
(1989) and Edvinsson et al. (1994). Bayer Technology Services, Bayer AG, Germany.

AIChE Journal November 2004 Vol. 50, No. 11 2935

Notation b ⫽ bubble
c ⫽ capillary, channel
a ⫽ interfacial area per unit volume, 1/L D ⫽ Darcy
C ⫽ factor F ⫽ Fanning
C ⫽ distribution parameter fr, f ⫽ frictional
cpsi ⫽ cells (or channels) per square inch g ⫽ gas
D ⫽ diffusion coefficient, m2/s G ⫽ gas
db ⫽ bubble diameter GL ⫽ gas–liquid
dc ⫽ capillary diameter GS ⫽ gas–solid
dh ⫽ hydraulic diameter m h ⫽ hydraulic
dP ⫽ dispersed bubble diameter, m l ⫽ liquid
dp/dz ⫽ pressure gradient, Pa/m L ⫽ liquid
DR ⫽ reactor diameter, m LF ⫽ liquid film alongside a Taylor bubble
F ⫽ factor LS ⫽ liquid–solid
f ⫽ friction factor m ⫽ measured
g ⫽ gravity, m/s2 orif ⫽ orifice
GSA ⫽ geometric surface area, in.2/in.3 P ⫽ dispersed bubble
k ⫽ mass transfer coefficient, m/s R ⫽ reactor
L ⫽ length, m s ⫽ slug
L ⫽ cell spacing, in. st ⫽ hydrostatic
Lb ⫽ gas bubble length, m tot ⫽ total
Lb ⫽ constant bubble length (dc ⫽ 2 mm), m TP ⫽ two phase
Lc ⫽ channel length z ⫽ length
LR ⫽ reactor length, m
Ls ⫽ liquid slug length, m
MR ⫽ monolith reactor Superscripts
n ⫽ cell density, cpsi
OFA ⫽ open frontal area * ⫽ dimensionless velocity
P/V ⫽ power input per unit volume of the system, W/m3 n ⫽ scaling factor
R ⫽ radius of the capillary, m
R ⫽ radius of gas plug, m
Re ⫽ Reynolds number ⫽ vd␳/␮ Literature Cited
Sc ⫽ Schmidt number ⫽ ␮␳/D
Sh ⫽ Sherwood number ⫽ kd/D Albers, R. E., M. Nyström, M. Siverström, A. Sellin, A. C. Dellve, U.
t ⫽ wall thickness, mm Andersson, H. Herrmann, and Th. Berglin, “Development of Monolith
TBR ⫽ trickle-bed reactor Based Process for H2O2 Production: From Idea to Large-Scale Imple-
UCL ⫽ unit cell length ⫽ Lb ⫹ Ls mentation,” Catal. Today, 69, 247 (2001).
uG ⫽ superficial gas velocity, m/s Al-Dahhan, M. H., and W. Highfill, “Liquid Holdup Measurement Tech-
uL ⫽ superficial liquid velocity, m/s niques in Laboratory High Pressure Trickle Bed Reactors,” Can.
uL,m ⫽ maximum liquid velocity, m/s J. Chem. Eng., 77(4), 759 (1999).
utot ⫽ sum of superficial gas and liquid velocity (uL ⫹ uG), m/s Al-Dahhan, M. H., F. Larachi, M. P. Dudukovic, and A. Laurent, “High-
V ⫽ volume, m3 Pressure Trickle-Bed Reactors: A Review,” Ind. Eng. Chem. Res., 36(8),
v ⫽ cell velocity, m/s 3292 (1997).
vb ⫽ bubble velocity, m/s Andersson, B., S. Irandoust, and A. Cybulski, “Modeling of Monolith
vG ⫽ gas velocity, m/s Reactors in Three-Phase Processes,” Chem. Ind. Struct. Catal. & Reac-
VG,u ⫽ drift velocity, m/s tors (Dekker), 71, 267 (1998).
We ⫽ Weber number ⫽ ␳lv 2l dp/␴ Bercić, G., and A. Pintar, “The Role of Gas Bubbles and Liquid Slug
x ⫽ quality Lengths on Mass Transport in the Taylor Flow through Capillaries,”
⌬p ⫽ pressure drop Chem. Eng. Sci., 52(21–22), 3709 (1997).
⌬P/⌬L ⫽ pressure drop per unit length, Pa/m Biswas, J., and P. F. Greenfield, “Two-Phase Flow through Vertical Cap-
⌬PG ⫽ pressure drop caused by gas, Pa illaries—Existence of a Stratified Flow Pattern,” Int. J. Multiphase Flow,
11(4), 553 (1985).
Boyer, C., and B. Fanget, “Measurement of Liquid Flow Distribution in
Greek letters Trickle Bed Reactor of Large Diameter with a New Gamma-Ray To-
mographic System,” Chem. Eng. Sci., 57, 1079 (2002).
␣ ⫽ dimensionless film thickness ⫽ ␦f/dt Broekhuis, R. R., R. M. Machado, and A. F. Nordquist, “The Ejector-
⌽ ⫽ volumetric flow rate, m3/s Driven Monolith Loop Reactor—Experiments and Modeling,” Catal.
⌿s ⫽ liquid slug length, m Today, 69(1– 4), 87 (2001).
␦f ⫽ liquid film thickness, m Chen, J., R. Novica, M. H. Al-Dahhan, and M. P. Dudukovic, “Study of
␦f,m ⫽ maximum film thickness, m Particle Motion in Packed/Ebullated Beds by Computed Tomography
␧ ⫽ bed void fraction (CT) and Computer Automated Radioactive Particle Tracking
␧G , ␤ ⫽ gas holdup (CARPT),” AIChE J., 47(5), 994 (2001).
␧gG ⫽ volume fraction of bubbles Coleman, J. W., and S. Garimella, “Characterization of Two-Phase Flow
␧gG/Lb ⫽ number of bubbles per unit length Patterns in Small Diameter Round and Rectangular Tubes,” Int. J. Heat
␧L ⫽ liquid holdup Mass Transfer, 42(15), 2869 (1999).
␥ ⫽ Lc/dc Crynes, L. L., R. L. Cerro, and M. A. Abraham, “Monolith Froth Reactor:
␮ ⫽ dynamic viscosity, kg/ms Development of a Novel Three-Phase Catalytic System,” AIChE J.,
␳ ⫽ density, kg/m3 41(2), 337– 45 (1995).
␴ ⫽ surface tension, N/m Cybulski, A., R. Edvinsson, S. Irandoust, and B. Andersson, “Liquid-Phase
␨ ⫽ dimensionless tube length Methanol Synthesis: Modelling of a Monolithic Reactor,” Chem. Eng.
Sci., 48(20), 3463 (1993).
Subscripts Cybulski, A., and J. A. Moulijn, Structured Catalysts and Reactors, Marcel
Dekker, New York (1998).
A ⫽ species A Cybulski, A., A. Stankiewicz, R. K. Edvinsson Albers, and J. A. Moulijn,
B ⫽ species B “Monolithic Reactors for Fine Chemicals Industries: A Comparative

2936 November 2004 Vol. 50, No. 11 AIChE Journal

 Analysis of a Monolithic Reactor and a Mechanically Agitated Slurry
Reactor,” Chem. Eng. Sci., 54(13–14), 2351 (1999).
Edvinsson, R. K., and A. Cybulski, “A Comparative Analysis of the
Trickle-Bed and the Monolithic Reactor for Three-Phase Hydrogena-
tions,” Chem. Eng. Sci., 49(24), 5653 (1994).
Edvinsson, R. K., A. M. Holmgren, and S. Irandoust, “Liquid-Phase
Hydrogenation of Acetylene in a Monolithic Catalyst Reactor,” Ind.
Eng. Chem. Res., 34(1), 94 (1995).
Edvinsson, R. K., M. J. J. Houterman, T. Vergunst, E. Grolman, and J. A.
Moulijn, “Novel Monolithic Stirred Reactor,” AIChE J., 44(11), 2459
(1998).
Edvinsson, R. K., and S. Irandoust, “Hydrodesulfurization of Dibenzothio-
phene in a Monolithic Catalyst Reactor,” Ind. Eng. Chem. Res., 32(2),
391 (1993).
Ellenberger, J., and R. Krishna, “Counter-Current Operation of Structured
Catalytically Packed Distillation Columns: Pressure Drop, Holdup and
Mixing,” Chem. Eng. Sci., 54(10), 1339 (1999).
Folger, H. S., Elements of Chemical Reaction Engineering, Prentice Hall,
Englewood Cliffs, NJ (1992).
Froment, G. B., and K. B. Bischoff, Chemical Reactor Analysis and
Design, Wiley, New York (1990).
Garcia-Bordeje, E., F. Kapteijn, and J. A. Moulijn, “Preparation and
Characterisation of Carbon-Coated Monoliths for Catalyst Supports,”
Carbon, 40(7), 1079 (2002).
Gulati, S. T., “Ceramic Catalyst Supports for Gasoline Fuel,” Structured
Catalysts and Reactors, A. Cybulski and J. A. Moulijn, eds., Marcel
Dekker, New York, pp. 15–58 (1998).
Harter, I., C. Boyer, L. Raynal, G. Ferschneider, and T. Gauthier, “Flow
Distribution Studies Applied to Deep Hydro-Desulfurization,” Ind. Eng.
Chem. Res., 40(23), 5262 (2001).
Hatziantoniou, V., and B. Andersson, “Solid–Liquid Mass Transfer in
Segmented Gas–Liquid Flow through a Capillary,” Ind. Eng. Chem.
Fundam., 21(4), 451 (1982).
Hatziantoniou, V., and B. Andersson, “The Segmented Two-Phase Flow
Monolithic Catalyst Reactor. An Alternative for Liquid-Phase Hydro-
genations,” Ind. Eng. Chem. Fundam., 23(1), 82 (1984).
Hatziantoniou, V., B. Andersson, and N. H. Schoon, “Mass Transfer and
Selectivity in Liquid-Phase Hydrogenation of Nitro Compounds in a
Monolithic Catalyst Reactor with Segmented Gas–Liquid Flow,” Ind.
Eng. Chem. Proc. Des. Dev., 25(4), 964 (1986).
Heibel, A. K., J. J. Heiszwolf, F. Kapteijn, and J. A. Moulijn, “Influence of
Channel Geometry on Hydrodynamics and Mass Transfer in the Mono-
lith Film Flow Reactor,” Catal. Today, 69(1– 4), 153 (2001a).
Heibel, A. K., T. W. J. Scheenen, J. J. Heiszwolf, H. Van As, F. Kapteijn,
and J. A. Moulijn, “Gas and Liquid Phase Distribution and Their Effect
on Reactor Performance in the Monolith Film Flow Reactor,” Chem.
Eng. Sci., 56(21–22), 5935 (2001b).
Heiszwolf, J. J., B. L. Engelvaart, G. M. van den Eijnden, T. M. Kreutzer,
F. Kapteijn, and J. A. Moulijn, “Hydrodynamic Aspects of the Monolith
Loop Reactor,” Chem. Eng. Sci., 56(3), 805 (2001a).
Heiszwolf, J. J., M. T. Kreutzer, M. G. van den Eijnden, F. Kapteijn, and
J. A. Moulijn, “Gas–Liquid Mass Transfer of Aqueous Taylor Flow in
Monoliths,” Catal. Today, 69(1– 4), 51 (2001b).
Highfill, W., and M. H. Al-Dahhan, “Liquid–Solid Mass Transfer Coeffi-
cient in High Pressure Trickle Bed Reactors,” Trans. IChemE, 79(A)
(2001).
Iliuta, I., and F. Larachi, “Mechanistic Model for Structured-Packing
Containing Columns Irrigated Pressure Drop, Liquid Holdup, and Pack-
ing Fractional Wetted Area,” Ind. Eng. Chem. Res., 40, 5140 (2000).
Irandoust, S., and B. Andersson, “Mass Transfer and Liquid-Phase Reac-
tions in a Segmented Two-Phase Flow Monolithic Catalyst Reactor,”
Chem. Eng. Sci., 43(8), 1983 (1988a).
Irandoust, S., and B. Andersson, “Monolithic Catalysts for Nonautomobile
Applications,” Catal. Rev. Sci. Eng., 30(3), 341 (1988b).
Irandoust, S., and B. Andersson, “Liquid Film in Taylor Flow through a
Capillary,” Ind. Eng. Chem. Res., 28(11), 1684 (1989a).
Irandoust, S., and B. Andersson, “Simulation of Flow and Mass Transfer in
Taylor Flow through a Capillary,” Comput. Chem. Eng., 13(4 –5), 519
(1989b).
Irandoust, S., B. Andersson, E. Bengtsson, and M. Siverstroem, “Scaling
Up of a Monolithic Catalyst Reactor with Two-Phase Flow,” Ind. Eng.
Chem. Res., 28(10), 1489 (1989c).
Irandoust, S., S. Ertle, and B. Andersson, “Gas–Liquid Mass Transfer in
AIChE Journal November 2004
Taylor Flow through a Capillary,” Can. J. Chem. Eng., 70(1), 115
(1992).
Irandoust, S., and O. Gahne, “Competitive Hydrodesulfurization and Hy-
drogenation in a Monolithic Reactor,” AIChE J., 36(5), 746 (1990).
Kapteijn, F., T. A. Nijhuis, J. J. Heiszwolf, and J. A. Moulijn, “New
Non-Traditional Multiphase Catalytic Reactors Based on Monolithic
Structures,” Catal. Today, 66(2– 4), 133 (2001).
Kawakami, K., K. Kawasaki, F. Shiraishi, and K. Kusunoki, “Performance
of a Honeycomb Monolith Bioreactor in a Gas–Liquid–Solid Three-
Phase System,” Ind. Eng. Chem. Res., 28(4), 394 (1989).
Klinghoffer, A. A., R. L. Cerro, and M. A. Abraham, “Catalytic Wet
Oxidation of Acetic Acid Using Platinum on Alumina Monolith Cata-
lyst,” Catal. Today, 40(1), 59 (1998a).
Klinghoffer, A. A., R. L. Cerro, and M. A. Abraham, “Influence of Flow
Properties on the Performance of the Monolith Froth Reactor for Cata-
lytic Wet Oxidation of Acetic Acid,” Ind. Eng. Chem. Res., 37(4), 1203
(1998b).
Kolb, W. B., and R. L. Cerro, “Coating the Inside of a Capillary of Square
Cross Section,” Chem. Eng. Sci., 46(9), 2181 (1991).
Kolb, W. B., and R. L. Cerro, “Film Flow in the Space between a Circular
Bubble and a Square Tube,” J. Colloid Interface Sci., 159(2), 302
(1993).
Kreutzer, M. T., P. Du, J. J. Heiszwolf, F. Kapteijn, and J. A. Moulijn,
“Mass Transfer Characteristics of Three-Phase Monolith Reactors,”
Chem. Eng. Sci., 56(21–22), 6015 (2001).
Krishna, R., and S. T. Sie, “Strategies for Multiphase Reactor Selection,”
Chem. Eng. Sci., 49(24A), 4029 (1994).
Kumar, S. B., M. P. Dudukovic, J. Chaouki, F. Larachi, and M. P.
Dudukovic, “Computer Assisted Gamma and X-ray Tomography: Ap-
plications to Multiphase Flow Systems,” Non-Invasive Monitoring of
Multiphase Flows, Elsevier, Amsterdam, pp. 47–103 (1997).
Lachman, I. M., and J. L. Williams, “Extruded Monolithic Catalyst Sup-
ports,” Catal. Today, 14(2), 317 (1992).
Lebens, P. J. M., J. J. Heiszwolf, F. Kapteijn, S. T. Sie, and J. A. Moulijn,
“Gas–Liquid Mass Transfer in an Internally Finned Monolith Operated
Countercurrently in the Film Flow Regime,” Chem. Eng. Sci., 54(21),
5119 (1999a).
Lebens, P. J. M., F. Kapteijn, S. T. Sie, and J. A. Moulijn, “Potentials of
Internally Finned Monoliths as a Packing for Multifunctional Reactors,”
Chem. Eng. Sci., 54(10), 1359 (1999b).
Lebens, P. J. M., M. M. Stork, F. Kapteijn, S. T. Sie, and J. A. Moulijn,
“Hydrodynamics and Mass Transfer Issues in a Countercurrent Gas–
Liquid Internally Finned Monolith Reactor,” Chem. Eng. Sci., 54(13–
14), 2381 (1999c).
Lebens, P. J. M., R. van der Meijden, R. K. Edvinsson, F. Kapteijn, S. T.
Sie, and J. A. Moulijn, “Hydrodynamics of Gas–Liquid Countercurrent
Flow in Internally Finned Monolithic Structures,” Chem. Eng. Sci.,
52(21–22), 3893 (1997).
Levenspiel, O., The Chemical Reactor Omnibook, OSU Book Stores, Inc.,
Corvallis, OR (1996).
Levenspiel, O., Chemical Reaction Engineering, Wiley, New York (1998).
Liu, W., “Ministructured Catalyst Bed for Gas–Liquid–Solid Multiphase
Catalytic Reaction,” AIChE J., 48(7), 1519 (2002).
Liu, W., W. P. Addiego, C. M. Sorensen, and T. Boger, “Monolith Reactor
for the Dehydrogenation of Ethylbenzene to Styrene,” Ind. Eng. Chem.
Res., 41(13), 3131 (2002).
Marcandelli, C., A. S. Lamine, J. R. Bernard, and G. Wild, “Liquid
Distribution in Trickle-Bed Reactor,” Oil Gas Sci Technol. Rev. IFP,
55(4), 407 (2000).
Mazzaroni, I., and G. Baldi, “Liquid-Phase Hydrogenation on a Monolith
Catalyst,” Recent Trends in Chemical Reaction Engineering, Vol. 2,
B. D. Kulkarni, R.A. Mashelkar, and M. M. Sharma, eds., Wiley Eastern
Ltd. New Delhi, p. 181 (1987).
Mewes, D., T. Loser, and M. Millies, “Modelling of Two-Phase Flow in
Packings and Monoliths,” Chem. Eng. Sci., 54(21), 4729 (1999).
Mishima, K., and T. Hibiki, “Some Characteristics of Air–Water Two-
Phase Flow in Small Diameter Vertical Tubes,” Int. J. Multiphase Flow,
22(4), 703 (1996).
Mishima, K., and M. Ishii, “Flow Regime Transition Criteria for Upward
Two-Phase Flow in Vertical Tubes,” Int. J. Heat Mass Transfer, 27(5),
723 (1984).
Mishra, V. S., V. V. Mahajani, and J. B. Joshi, “Wet Air Oxidation,” Ind.
Eng. Chem. Res., 34(1), 2 (1995).
Nemec, D., G. Bercić, and J. Levec, “Gravimetric Method for the Deter-
Vol. 50, No. 11 2937
 mination of Liquid Holdup in Pressurized Trickle-Bed Reactors,” Ind.
Eng. Chem. Res., 40(15), 3418 (2001).
Nijhuis, T. A., F. M. Dautzenberg, and J. A. Moulijn, “Modeling of
Monolithic and Trickle-Bed Reactors for the Hydrogenation of Styrene,”
Chem. Eng. Sci., 58, 1113 (2003).
Nijhuis, T. A., M. T. Kreutzer, A. C. J. Romijn, F. Kapteijn, and J. A.
Moulijn, “Monolithic Catalysts as Efficient Three-Phase Reactors,”
Chem. Eng. Sci., 56(3), 823 (2001a).
Nijhuis, T. A., M. T. Kreutzer, A. C. J. Romijn, F. Kapteijn, and J. A.
Moulijn, “Monolithic Catalysts as More Efficient Three-Phase Reac-
tors,” Catal. Today, 66(2– 4), 157 (2001b).
Patrick, R. H., Jr., T. Klindera, L. L. Crynes, R. L. Cerro, and M. A.
Abraham, “Residence Time Distribution in Three-Phase Monolith Re-
actor,” AIChE J., 41(3), 649 (1995).
Patrick, T. A., and M. A. Abraham, “Evaluation of a Monolith-Supported
Pt/Al2O3 Catalyst for Wet Oxidation of Carbohydrate-Containing Waste
Streams,” Environ. Sci. Technol., 34, 3480 (2000).
Pinna, D., E. Tronconi, and L. Tagliabue, “High Interaction Regime
Lockhart–Martinelli Model for Pressure Drop in Trickle-Bed Reactors,”
AIChE J., 47(1), 19 (2001).
Ramachandran, P. A., and R. V. Chaudhari, Three-Phase Catalytic Reac-
tors, Gordon & Breach, New York (1983).
Reinecke, N., and D. Mewes, “Flow Regimes of Two Phase Flow in
Monolith Catalyst,” Proc. of the 5th World Congress on Chemical
Engineering, San Diego, CA (1996).
Reinecke, N., and D. Mewes, “Oscillatory Transient Two-Phase Flows in
Single Channels with Reference to Monolithic Catalyst Supports,” Int. J.
Multiphase Flow, 25(6 –7), 1373 (1999).
Reinecke, N., G. Petritsch, D. Schmitz, and D. Mewes, “Tomographic
Measurement Techniques—Visualization of Multiphase Flows,” Chem.
Eng. Technol., 21, 7 (1998).
Roy, S., A. K. Heibel, W. Liu, and T. Boger, “Design of Monolithic
Catalysts for Multiphase Reactions,” Oral Presentation, 17th SCRE
(2002).
Satterfield, C. N., and F. Ozel, “Some Characteristics of Two-Phase Flow
in Monolithic Catalyst Structures,” Ind. Eng. Chem. Fundam., 16(1), 61
(1977).
Schluter, V., and W. D. Deckwer, “Gas–Liquid Mass Transfer in Stirred
Vessels,” Chem. Eng. Sci., 47, 2357 (1992).
Schöön, N. H., “Recent Progress in Liquid-Phase Hydrogenation: With
2938 November 2004
Aspects from Microkinetics to Reactor Design,” Proc. of the 6th Natl.
Sympos. Chem. React. Eng., Warsaw, Poland (1989).
Schutt, B. D., B. Serrano, R. L. Cerro, and M. A. Abraham, “Production of
Chemicals from Cellulose and Biomass-Derived Compounds through
Catalytic Sub-Critical Water Oxidation in a Monolith Reactor,” Biomass
& Bioenergy, 22(5), 365 (2002).
Smits, H. A., A. Stankiewicz, W. C. Glasz, T. H. A. Fogl, and J. A.
Moulijn, “Selective Three-Phase Hydrogenation of Unsaturated Hydro-
carbons in a Monolithic Reactor,” Chem. Eng. Sci., 51(11), 3019 (1996).
Stankiewicz, A., “Process Intensification in In-Line Monolithic Reactor,”
Chem. Eng. Sci., 56(2), 359 (2001).
Stankiewicz, A. I., and J. A. Moulijn, “Process Intensification: Transform-
ing Chemical Engineering,” Chem. Eng. Prog., 96(1), 22 (2000).
Thulasidas, T. C., M. A. Abraham, and R. L. Cerro, “Bubble-Train Flow in
Capillaries of Circular and Square Cross Section,” Chem. Eng. Sci.,
50(2), 183 (1995a).
Thulasidas, T. C., M. A. Abraham, and R. L. Cerro, “Dispersion during
Bubble-Train Flow in Capillaries,” Chem. Eng. Sci., 54(1), 61 (1999).
Thulasidas, T. C., R. L. Cerro, and M. A. Abraham, “The Monolith Froth
Reactor: Residence Time Modeling and Analysis,” Chem. Eng. Res.
Des., 73(A3), 314 (1995b).
Triplett, K. A., S. M. Ghiaasiaan, S. I. Abdel-Khalik, A. LeMouel, and
B. N. McCord, “Gas–Liquid Two-Phase Flow in Microchannels. Part II:
Void Fraction and Pressure Drop,” Int. J. Multiphase Flow, 25(3), 395
(1999a).
Triplett, K. A., S. M. Ghiaasiaan, S. I. Abdel-Khalik, and D. L. Sadowski,
“Gas–Liquid Two-Phase Flow in Microchannels. Part I: Two-Phase
Flow Patterns,” Int. J. Multiphase Flow, 25(3), 377 (1999b).
Van’t Riet, K., “Review of Measuring Methods and Results in Nonviscous
Gas–Liquid Mass Transfer in Stirred Vessels,” Ind. Eng. Chem. Proc.
Dev., 18, 357 (1979).
Williams, J. L., “Monolith Structures, Materials, Properties and Uses,”
Catal. Today, 69(1– 4), 3 (2001).
Woehl, P., and R. L. Cerro, “Pressure Drop in Monolith Reactors,” Catal.
Today, 69(1– 4), 171 (2001).
Zhao, T. S., and Q. C. Bi, “Co-Current Air–Water Two-Phase Flow
Patterns in Vertical Triangular Microchannels,” Int. J. Multiphase Flow,
27(5), 765 (2001).
Manuscript received July 14, 2003, and revision received Feb. 29, 2004.
Vol. 50, No. 11 AIChE Journal